Design, Synthesis, and Evaluation of Rhodol-Based Fluorescent Copper Sensors Copper Rhodol-3 and CF3.

We previously used X-ray fluorescence microscopy for direct total copper detection, inspired by initial pilot data using the BODIPY (boron dipyrromethene)-based fluorescent probe Coppersensor-3 (CS3) for labile Cu⁺ imaging, to identify that dissociated cultured neurons activated by exogenous chemical depolarization agents mobilize pools of copper from their somatic cell bodies to peripheral processes in a calcium-dependent manner (29). Encouraged by these findings, we sought to probe loosely bound copper pools in higher neuronal tissue models and study their roles in regulating neural circuits under basal conditions with a molecular imaging approach. However, initial attempts to use CS3 and other available copper indicators to reliably visualize labile copper stores in tissue were unsuccessful, despite their utility in a wide range of cell culture models (29⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓–41). We speculated that the relative hydrophobicity of BODIPY dyes could potentially limit their utility in thicker biological specimens due to localization and aggregation effects (30, 42, 43). This situation is unsurprising, as the invention of fluorescent probes is a developing technology and not every probe can be used with the same efficacy in every cell type. In particular, the requirement for careful synthesis, purification, and handling of CS3 due to its relatively short shelf life, even in solid form (41), led us to explore alternative sensor platforms for copper detection. To address this technological gap, we designed and synthesized a family of fluorescent copper sensors, Copper Rhodols 1–5 (CR1−5), based on the rhodol scaffold as shown in Fig. 1A. Rhodols are hybrid fluorescein−rhodamine dyes that offer improved hydrophilicity over their BODIPY counterparts yet maintain good optical brightness, photostability, and insensitivity to changes in the physiological pH range (44). These properties allow them to be used for both one- and two-photon microscopy, where the latter technique expands their utility for imaging tissue and other thicker specimens (45).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Preparation and characterization of CR3 and CF3. (A) Schematic of BODIPY- and rhodol-based copper sensors. (B) Synthetic scheme of CR3 and CF3.

Beginning with the most responsive sensor of the series, Copper Rhodol-3 (CR3), we strategically made a methyl to trifluoromethyl substitution on the bottom ring receptor to afford CF3, which exhibits enhanced photophysical properties (vide infra). Moreover, a set of matched control rhodols, Control Copper Rhodol-3 (Ctrl-CR3) and Ctrl-CF3, that lack a copper-responsive receptor were prepared to distinguish between receptor-dependent and dye-dependent responses. Specifically, the thioether-rich receptor arms for Cu⁺ recognition in CR3 and CF3 were replaced by isostructural octyl groups in Ctrl-CR3 and Ctrl-CF3 that mimic the size, shape, and hydrophobicity of thioethers but do not bind copper. The syntheses of the fluorescent copper indicators, CR3 and CF3, as well as the control analogs, Ctrl-CR3 and Ctrl-CF3, are depicted in Fig. 1B and Fig. S1A. The preparation and characterization of all other copper sensors are described in SI Materials and Methods. The general syntheses proceeded through addition of the aryl lithium species, derived from the bromoarene via lithium halogen exchange, to the appropriate xanthone. Briefly, xanthone 3 was accessed in two steps starting from the nucleophilic aromatic substitution of 1 with stoichiometric pyrrolidine to give the monopyrrolidino xanthone 2. Subsequent one-pot conversion to xanthone 3 involved hydrolysis of the triflate with tetraethylammonium hydroxide, followed by protection of the phenol with tert-butyldimethylsilyl chloride. The bromobenzene precursors 8 and 12 were prepared from the quantitative reduction of methyl 4-bromo-3-methylbenzoate 4 and methyl 4-bromo-3-(trifluoromethyl)benzoate 9 with lithium aluminum hydride to afford benzyl alcohols 5 and 10, respectively. The benzyl chlorides 6 and 11 were accessed, respectively, by refluxing 5 and 10 in solutions of thionyl chloride in dichloromethane. Finally, incorporation of the thioether-rich binding domain via nucleophilic displacement of the benzyl chloride with bis(2-((2-(ethylthio)ethyl)thio)ethyl)amine 7 afforded bromobenzenes 8 and 12, respectively. Likewise, substitution with dioctylamine allowed for the preparation of the nonresponsive copper control precursors 14 and 15, respectively. The appropriate xanthone and bromobenzene synthons were then coupled to give the final CR3 and CF3 probes and their control compounds.

We reasoned that replacement of the methyl substituent on the bottom ring receptor of CR3 with a trifluoromethyl group would result in enhancements in dynamic range and optical brightness by (i) providing greater steric bulk to minimize nonradiative decay by rotational motions between the receptor-containing aryl and xanthene rings and (ii) enhancing quenching of the unbound state by photoinduced electron transfer and/or alleviating quenching related to charge-transfer mechanisms by decreasing electron density on the receptor-containing aryl ring. Indeed, CR3 gives a 13-fold turn-on fluorescence enhancement upon binding to Cu⁺, but CF3 responds to Cu⁺ with a 40-fold emission increase (Fig. 2 A and C) with a 20-nM detection limit (Fig. S3E). Both sensors exhibit high selectivity for Cu⁺, even in the presence of physiologically relevant concentrations of competing metal ions (Fig. 2 B and D). The fluorescence responses of CR3 and CF3 to Cu⁺ are not affected by the presence of 2 mM Ca²⁺, Mg²⁺, and Zn²⁺, nor do these metal ions cause increases in observed fluorescence signals. Moreover, other biologically abundant transition metal ions (50 µM Mn²⁺, Fe²⁺, Co²⁺, Ni²⁺, or Cu²⁺) do not trigger false positives nor do they interfere with Cu⁺-induced fluorescence enhancements for CR3 and CF3. Notably, Cu²⁺ data show that both indicators maintain oxidation state specificity for Cu⁺ over Cu²⁺. Additionally, the Ctrl-CR3 and Ctrl-CF3 probes do not respond to Cu⁺ or any other metal ions (Table S1). Finally, based on determination of partition coefficients, we find that CR3 and CF3 are more hydrophilic (log D = 0.96 for CR3, 1.15 for CF3) compared with the BODIPY counterpart CS3 (log D = 3.46), potentially enabling the broader use of these CR3/CF3 platforms in both cell and tissue specimens (Table S1). Further in vitro experiments establish that CF3, but not Ctrl-CF3, is able to respond to copper in the presence of artificial lipid vesicles (Fig. S3 F−I), BSA as a model protein (Fig. S3J), and millimolar levels of glutathione (GSH) (Fig. S3 K and L), an abundant competing cellular ligand for Cu(I) (apparent K_d = 9 × 10⁻¹² assuming the major species is a 1:1 Cu:GSH complex) (46, 47), as well as cell lysates that contain all of these components (Fig. S3M).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Plot of fluorescence increase of 2 µM (A) CR3 and (C) CF3 to Cu⁺. Points are shown for 0 µM, 0.4 µM, 0.8 µM, 1.2 µM, 1.6 µM, and 2 µM Cu⁺. Excitation was provided at 534 nm and collected from 545 to 700 nm. Fluorescence responses of (B) CR3 and (D) CF3 to various metal ions. Bars represent the final integrated fluorescence response (F_f) over the initial integrated emission (F_i). White bars represent the addition of an excess of the indicated metal ion (2 mM for Na⁺, K⁺, Mg²⁺, Ca²⁺, and Zn²⁺; 50 μM for all other cations) to a 2 μM solution of CR3 or CF3. Black bars represent subsequent addition of 2 μM Cu⁺ to the solution. Excitation was provided at 534 nm, and collected emission was integrated from 545 nm to 700 nm.

Molecular Imaging with CF3 Revealed Pools of Loosely Bound Copper in Cultured Dissociated Hippocampal Neurons and Acutely Isolated Retinal Tissue.

The enhanced sensitivity and hydrophilicity of CF3 enabled the study of exchangeable Cu⁺ stores in neuronal circuits formed from dissociated hippocampal cultures and retinal tissue models using confocal or two-photon scanning imaging (Fig. 3 and Fig. S8A). Dissociated hippocampal neurons (days in vitro (DIV) 10–13) loaded with CF3 for 20 min followed by perfusion with buffered saline [Hank’s Balanced Salt Solution (HBSS)] (see SI Materials and Methods) and imaging with two-photon confocal microscopy showed a diffuse intracellular fluorescent signal (Fig. 3A). Acute bath application of the cell-impermeable copper chelator bathocuproine disulfonate (BCS) to the same field of neurons showed a patent decrease in intracellular fluorescence intensity. The data suggest that CF3 is capable of detecting basal, endogenous pools of labile copper in neurons and their depletion by on-stage pharmacological treatments within minutes of chelation (Fig. 3B), as BCS has a higher affinity for Cu⁺ compared with CF3 (1 × 10⁻²⁰ vs. 3.3 × 10⁻¹³) and therefore alters intracellular pools of loosely bound copper pools by sequestering extracellular copper stores (48). Notably, the application of Ctrl-CF3, which possesses an identical fluorophore but lacks the copper responsive binding element, under the same experimental conditions showed no change in fluorescence intensity upon treatment with BCS. This critical control experiment confirmed that the observed BCS-induced decreases in CF3 fluorescence were not due to photobleaching and/or dye leakage effects. Moreover, BCS was not observed to induce general oxidative stress in hippocampal cultures (Fig. S8C), further supporting that observed changes in CF3 intensity are copper dependent. Because GSH is an abundant species in cells and a potential competing ligand for labile copper, we speculate that CF3 could exchange or form a ternary complex with GSH-bound copper (46, 47). Indeed, intracellular CF3 fluorescence in hippocampal neurons decreased upon treatment with the cell-permeable copper chelator GSH monoethyl ester (GSH-MEE) (Fig. S8B), suggesting that CF3 can reversibly respond to changes in the GSH-dependent copper pool.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Live two-photon imaging of CF3 in cultured dissociated hippocampal neurons and retinal tissue shows that acute BCS treatment alters intracellular labile Cu⁺. Representative images are shown of hippocampal neurons were stained with (A) 2 μM CF3 (CF3: baseline normalized mean = 0.32, SD = 0.15; after BCS: mean = 0.088, SD = 0.038; n = 192 15-μm squares) or (B) Ctrl-CF3 for 20 min with subsequent recording of baseline fluorescence followed by 30 min BCS perfusion (Ctrl-CF3: baseline normalized mean = 0.20, SD = 0.15; after BCS: mean = 0.20, SD = 0.15; n = 192 15-μm squares). Net changes in fluorescence brightness are quantified on the right with paired lines representing nonoverlapping 15-μM square regions of the field of view. Representative images of retinal neurons bolus loaded with (C) CF3 (CF3: baseline normalized mean = 0.49, SD = 0.16; after BCS: mean = 0.26, SD = 0.10; n = 256 15-μm squares) or (D) Ctrl-CF3 followed by 30 min BCS perfusion (Ctrl-CF3: baseline normalized mean = 0.57, SD = 0.13; after BCS: mean = 0.56, SD = 0.11; n = 256 15-μm squares). Net changes in fluorescence brightness are quantified on the right, with paired lines representing nonoverlapping 15-μm square regions of the field of view.

Next, acutely isolated retinal tissues (P10−P13) were loaded with CF3 and the fluorescence intensity was imaged from the ganglion cell layer (GCL) using a scanning two-photon microscope (Fig. S9A and Fig. 3C). The retina consists of distinct cellular layers that contain neuronal somata and are separated by two plexiform layers that contain the processes and synapses between these neurons. CF3 labeled the intracellular space of the innermost cell layer, the GCL, and responded to a decrease in the pool of loosely bound Cu⁺ upon perfusion with BCS (Fig. 3C). Similarly, loading retinas with Ctrl-CF3 followed by extracellular copper chelation did not result in a change in fluorescence, indicating that two-photon imaging of CF3 can be used to report on an endogenous pool of loosely bound Cu⁺ (Fig. 3D). Taken together, the data suggest that CF3 enables identification of loosely bound copper pools at endogenous levels in both dissociated neuronal cultures and in brain tissue, and that the membrane-impermeable copper chelator BCS alters the extracellular pool of loosely bound Cu⁺, which subsequently depletes the intracellular pool. As such, we reasoned that dynamic disruptions in loosely bound Cu⁺ stores may also result in alterations to spontaneous activity in neural networks.

Modulation of Spontaneous Activity with Acute Extracellular Copper Chelation in Dissociated Hippocampal Neurons.

First, we examined the effects of BCS treatment on spontaneous activity of a network of cultured hippocampal neurons using both confocal and two-photon based calcium imaging. Cultures of embryonic neurons (DIV 12–15) were incubated with the Ca²⁺ indicator, Oregon Green BAPTA-1 AM (OGB). These dissociated hippocampal cultures, which consist of neurons and glia, exhibited temporally uncorrelated calcium transients (Fig. 4A). Bath application of the copper chelator BCS in a dose-dependent manner (5−200 µM) triggered increases in the correlations between calcium transients (Fig. S8D and Fig. 4B). Importantly, uncorrelated calcium transients were restored upon washout of BCS, showing that this modulation is reversible (Fig. S8E). To compare quantitatively the correlated events, we computed cross-correlations based upon the percentage of imaged cells that exhibited spontaneous calcium transients surrounding the peak event (see SI Materials and Methods). In control conditions, a lower percentage of cells exhibited correlated calcium transients (Fig. 4B). These data are consistent with the model that disrupting loosely bound copper stores by acute chelation increases the excitability of the network, leading to a reversible increase in correlated activity (49).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Copper chelation with BCS increases the correlated calcium transients in dissociated hippocampal culture. (A) Sample ∆F/F traces from averaged regions from four different cells stained with OGB. Each trace row corresponds to the same in cell in control solution first (Left) followed by chelation with 200 µM BCS (Right). Binary plots (Bottom) show entire network activity, where a black point corresponds to a cell with a calcium transient above threshold. (B) Mean and SD across cells of cross-correlation during times of high-synchronized firing (n = 3 cultures). When a time point was identified as having more than 10% of cells above threshold, the proportion of cells that were above threshold was calculated for 5 s before and after. A flat line indicates uniform, random transients. WT (black) and acute 200 µM BCS wash-in (red).

Pharmacological Manipulation of Endogenous Pools of Loosely Bound Copper Alters Spontaneous Activity in the Developing Retina.

Next, we studied spontaneous activity in the developing retina, a well-understood model for correlated spontaneous activity during development (50, 51). Before the maturation of the light responses, the retina exhibits spontaneous correlated activity, termed retinal waves. To investigate the effects of copper chelation on retinal waves, we used two-photon calcium imaging to record from the neurons in the GCL in which resides a pool of exchangeable copper that was altered by acute BCS treatment (Fig. 5). Retinas were acutely isolated from P10−P13 mice and bolus loaded with OGB as previously described (52). Spontaneous calcium transients propagated across the GCL with a well-defined periodic frequency (Fig. 5 A and B). Acute bath application of 200 µM BCS increased the frequency of waves; the median interevent interval decreased for BCS-treated retinas (median, upper/lower quartile; control: 100 s, 60/130, n = 17 intervals; BCS: 29 s, 17/58 s, n = 51 intervals; wash wave interval: 60 s, 17/58, n = 6 intervals) (Fig. 5 C and D). In addition, BCS increased the percentage of GCL neurons that depolarized during correlated activity and this effect was reversible upon BCS washout (median, upper/lower quartile; control: 26%, 8.9/34; BCS: 75%, 56/78; wash: 12%, 1/20, n = 9 retinas; Kruskal−Wallis test, P < 0.001) (Fig. 5E). To confirm that BCS is functioning by altering labile copper concentrations, we applied ATN-224, a membrane-permeable copper chelator that is chemically distinct from BCS and irreversibly alters both intracellular and extracellular copper stores. Consistent with the BCS results, bath application of ATN-224 (15 μM) increased wave frequency in the mouse retina (median, upper/lower quartile; control: 153 s, 82/168, n = 7 intervals; ATN-224: 34 s, 16/74, n = 13 intervals; Wilcoxon signed-rank test, P < 0.05; Fig. S9D). BCS was the primary copper chelator used because of its cell-impermeable nature, which allows for reversible application to tissue and offers a better mimic of a genetic knockout of the membrane-bound CTR1 protein that controls copper entry into cells from the extracellular face. Similar to what we observed in the hippocampal cultures, the disruption of endogenous pools of loosely bound copper by acute addition of chelators increases excitability of retinal circuits, suggesting a role for copper in regulating spontaneous activity.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Copper chelation increases wave frequency and cell participation levels of correlated spontaneous activity in the developing retina. (A) Example of retinal wave propagation in the GCL observed with two-photon calcium imaging at a frame rat of 1 Hz. Leftmost image is the retinal sample loaded with OGB. Circles are identified cells where black indicates cells with above-threshold ∆F/F in that frame. (B) Sample ∆F/F traces from averaged regions within four different cells. Each trace row corresponds to the same cell in control solution first (Left) followed by chelation with 200 µM BCS (Right). (C) Binary plots of neuronal calcium transients exceeding threshold for all cells in the field of view. (D) Summary of effects of BCS on interevent interval where lines connect values of average wave interval for one retina in control versus chelation with BCS. Open circles are group means, and bars are SD. (E) Summary of effects of BCS on proportion of cells that are active during waves (n = 9 retinas, P < 0.001, Kruskal−Wallis test). Box boundaries indicate upper and lower quartiles, with the median plotted in red with whiskers extending to the most extreme values that are still considered not to be outliers.

Spontaneous Activity Is Altered in CTR1^+/− Retinal Tissue.

To probe the effects of genetic, long-term disruption of total copper in the developing retina, as well as provide a molecular target for dynamic regulation of copper cycling, we compared our findings from WT mice to retinas isolated from transgenic mice that have heterozygote expression of the CTR1 copper ion channel (CTR1^+/−), which has ∼50% less copper in the nervous system compared with WT CTR1^+/+ mice; we note that the homozygous CTR1^−/− mice are embryonic lethal and cannot be used for such studies (53). First, we tested the total copper content in eyes isolated from WT and CTR1^+/− by inductively coupled plasma mass spectrometry (ICP-MS). Eyes isolated from CTR1^+/− mice accumulated less copper compared with WT (Cu/S with SD; CTR1^+/+ Cu/S = 0.00030 ± 0.00003; CTR1^+/− Cu/S = 0.00025 ± 0.00002; n ≥ 6 eyes). We also confirmed that there was a patent reduction in CTR1 expression in CTR1^+/− mice retinas by Western blot (Fig. 6F). The direct ICP-MS measurements are supported by secondary imaging experiments, where CTR1^+/− retinas loaded with CF3 show lower levels of fluorescence compared with WT retinas stained with this dye under identical conditions, whereas WT and CTR1^+/− retinas loaded with the Ctrl-CF3 dye that does not respond to copper show comparable fluorescence signals (Fig. S9C). Retinas from P11 CTR1^+/− mice exhibited spontaneous calcium transients as observed in WT mice (Fig. 6 A and B). However, similar to what was observed for BCS-treated WT retinas, waves in CTR1^+/− retinas were more frequent than littermate controls (median interevent interval upper/lower quartile; CTR1^+/+: 95 s, 76/149, n = 33 intervals; CTR1^+/−: 22 s, 14/32, n = 129 intervals; Wilcoxon signed-rank test, P < 0.001) and the percentage of depolarized cells increased (median, upper/lower quartile; control WT: 27%, 16/51, n = 18 retinas; CTR1^+/−: 67%, 17/68, n = 16 retinas; Wilcoxon signed-rank test, P < 0.001) (Fig. 6 D and E). Taken together, the data establish that disruptions of endogenous copper pools by genetic knockdown of the copper ion channel CTR1 or acute treatment with chelators like BCS or ATN-224 increase network excitability in both hippocampal culture and intact retinas. The observed changes to spontaneous neuronal activity in both systems suggest a general, fundamental role for copper in modulating spontaneous activity in developing neural circuits and identify CTR1 as a molecular target for mediating this effect through maintenance of extracellular and intracellular pools of loosely bound copper.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Partial knockdown of CTR1 in the retina increases wave frequency and cell participation in retinal waves. (A) Example of wave propagation in the GCL observed with two-photon calcium imaging at a frame rate of 1 Hz. Leftmost image is the retinal sample loaded with OGB. Circles are identified cells where black indicates cells with ∆F/F greater than 15% in that frame. (B) Sample ∆F/F traces from averaged regions within four different cells. Each trace row represents a different cell. (C) Binary plots of neuronal calcium transients greater than 15% ∆F/F for all cells in the field of view. (D) Summary of differences between WT and CTR1^+/− retina on interevent interval (WT: n = 33 waves; CTR1^+/−: n = 129 waves). (E) Summary of differences between WT and CTR1^+/− retina on proportion of cells that are active during waves (WT: n = 18 retinas; CTR1^+/−: n = 16 retinas). Box boundaries indicate upper and lower quartiles, with the median plotted in red with whiskers extending to the most extreme values that are still considered not to be outliers. (F) Representative Western blot of CTR1 expression in WT and CTR1^+/− developing retina. GAPDH was used as a loading control.