Rhodol-based Thallium Sensors for Cellular Imaging of Potassium Channel Activity

Abstract

Thallium (Tl⁺) flux assays enable imaging of potassium (K⁺) channel activity in cells and tissues by exploiting the permeability of K⁺ channels to Tl⁺ coupled with a fluorescent Tl⁺ sensitive dye. Common Tl⁺ sensing dyes utilize fluorescein as the fluorophore though fluorescein exhibits certain undesirable properties in these assays including short excitation wavelengths and pH sensitivity. To overcome these drawbacks, the replacement of fluorescein with rhodols was investigated. A library of 13 rhodol-based Tl⁺ sensors was synthesized and their properties and performance in Tl⁺ flux assays evaluated. The dimethyl rhodol Tl⁺ sensor emerged as the best of the series and performed comparably to fluorescein-based sensors while demonstrating greater pH tolerance in the physiological range and excitation and emission spectra 30 nm red-shifted from fluorescein.

A rhodol-based thallium sensor improves upon currently used fluorescein-based sensors for imaging potassium channel activity.

Ion channels are pore-forming membrane proteins which promote the passage of ions across cellular membranes. Within the superfamily of ion channels are subfamilies with different ionic selectivity.¹ The largest subclass is potassium (K⁺) channels which are expressed in every tissue in the body where they participate in control of excitability of neurons, muscles, endocrine glands, and the heart^2–4 while in organs like the kidney they play critical roles in the formation of urine and maintenance of proper Na⁺/K⁺ homeostasis.⁵ K⁺ channel activity can be modulated by changes in transmembrane potential as well as a wide variety of extracellular and intracellular factors including Ca²⁺, membrane lipids, G-protein-coupled receptors, protein kinases, and ATP.⁶ A rapidly expanding number of mutations are being identified in K⁺ channels which can result in neurological, cardiovascular, renal, neoplastic, and metabolic diseases.^7,8,9 The physiological roles of many of these K⁺ channels remain uncharacterized underscoring the need for robust chemical tools to study them.

Electrophysiology is the most sensitive technique for assaying ion channel activity though it can be time consuming and technically challenging. An alternate approach is the use of ion flux assays where channel activity is monitored by changes in fluorescence of an ion-sensing dye as a measure of passage of that ion through the channel of interest. In particular, the thallium (Tl⁺) flux assay¹⁰ (Figure 1) is extensively used for imaging K⁺ channel activity in cells and tissue samples. Tl⁺ flux assays are highly compatible with kinetic imaging-based high-throughput screening (HTS) assays. HTS is an important method for the discovery of K⁺ channel modulators by enabling the rapid testing of thousands-to-millions of molecules to identify those with activity against the channel of interest. Tl⁺ flux has been used in numerous HTS campaigns to identify modulators of K⁺ channels implicated in disease^11–13 and have also been used to screen compounds for activity against K⁺ transporters, Na⁺ channels, and non-selective cation channels.^14,15

Figure 1.

Overview of Tl⁺ flux assay.

Tl⁺ flux assays exploit the ability of Tl⁺ to pass through K⁺ channels coupled with a cell-permeable fluorescent Tl⁺ indicator. Several fluorescent Tl⁺ indicators have been reported with the most robust, such as Thallos (1), consisting of a fluorescein moiety coupled to an amino dicarboxylic acid metal-binding unit.¹⁶ The fluorescence pathway is partially quenched in the non-Tl⁺ bound form by a presumed photoinduced electron transfer (PET) between the metal-binding unit and fluorophore. When Tl⁺ is bound, the PET-pathway is inhibited resulting in a higher fluorescence quantum yield for the complex which is observed as a Tl⁺ concentration-dependent increase in fluorescence.

Fluorescein-based metal sensors have found great utility in biology as imaging agents and components of activity assays.¹⁷ However, there are drawbacks to their use including relatively short excitation wavelengths and an undesirable pH sensitivity in the physiological range leading to decreased brightness at lower pH.¹⁸ The short excitation wavelength is problematic in HTS assays since many compound libraries contain molecules with comparable excitation and emission spectra which interfere with the assay. The number of compounds in a library generating this kind of interference tends to decrease as excitation wavelength increases.¹⁹ In addition, autofluorescence in tissues tends to decrease when longer wavelengths are used for imaging using fluorescence microscopy.²⁰ Rhodamines, which are red-shifted ~50 nm relative to fluorescein and exhibit excellent pH tolerance, have also been used as the fluorescent components of metal sensors¹⁷, but have also been shown to accumulate in the mitochondria²¹ which is generally undesirable in flux assays that are focused on plasma membrane channels. Rhodols are a hybrid of fluorescein and rhodamines exhibiting excitation and emission spectra ~30 nm red-shifted from fluorescein, excellent pH tolerance in the physiological range²², and predominantly cytoplasmic accumulation.²³ PET-based metal sensors utilizing rhodol fluorophores have been reported for imaging Cu⁺²⁴, Na⁺²⁵, and Ca²⁺²⁶. Here we report the synthesis and evaluation of a library of rhodol-based Tl⁺ sensors for Tl⁺ flux assays of K⁺ channels and other Tl⁺ conducting systems.

A variable route to the rhodol component of the Tl⁺ sensor was envisioned utilizing a similar strategy as previously described²⁷ but incorporating protecting groups to allow for orthogonal deprotection of the phenol and distal carboxylic acid functional groups (Scheme 1). Starting from 5- carboxyfluorescein (2), mono-phenol protection as the benzyl ether was accomplished by treatment with benzyl chloride and potassium carbonate in DMF under microwave heating followed by benzyl ester saponification with lithium hydroxide to afford 3. Carboxylic acid 3 was then protected as the t-butyl ester and converted to triflate 4. From this common intermediate, rhodols 5a – m were prepared by Buchwald-Hartwig coupling with the corresponding amine. These compounds were converted to O-acetyl rhodols 6a – m by sequential hydrogenolysis of the benzyl ether, removal of the t-butyl ester by treatment with TFA, and O-acetylation. 6a and 6b were prepared from a slightly modified route (Supplemental Information). Carboxylic acids 6a – m were coupled to 7 to provide rhodol Tl⁺ sensors 8a – m. (Table 1)

Scheme 1.

Synthesis of rhodol Tl⁺sensors; a. BnCl, K₂CO₃, DMF, 120 °C, 1 h; b. 23LiOH, THF/H₂O, 60 °C,1 h, 59 % (2 steps); c. Ac₂ O pyridine, rt, 2 h, d., O-tert-butyldiisopropylisourea, DCM, rt, 48 h, then NaOMe, MeOH, rt, 20 min, 68 % (2 steps), e. Tf₂O, pyridine, DCM, 0 °C, 1 h, 78 %; f. amine, PdOAc₂, R-BINAP, Cs₂CO₃, PhMe, 100 °C, 20 h, 18 −90 %; g. H₂, Pd/C, EtOAc, rt, 1 h; h. TFA, DCM, rt, 12 h; i. Ac₂O, pyridine, rt, 2 h; 22 −67 % (3 steps); j. HATU, Et₃N, DMSO, rt, 1 h, 33 – 58 %.

Table 1.

Photochemical properties of rhodol Tl⁺sensors 8a-m.^a,b

Cmpd	X	λmax	λem	ε (M−1 * cm−1)	on/off (Tl+ flux assay)
8a	--NH2	496	520	57000	3.02 ± 0.08
8b	--NHAc	463	NA	24000	NA
8c	--NMe2	521	551	56000	4.21 ± 0.04
8d	- -NEt2	525	551	55000	3.28 ± 0.04
8e	- -NPr2	527	553	66000	2.94 ± 0.08
8f	--NBu2	528	556	58000	1.52 ± 0.01
8g		514	535	57000	4.26 ± 0.07
8h	--NHBu	511	533	48000	3.50 ± 0.07
8i	--NHPh	517	NA	52000	1.07 ± 0.01
8j		522	550	56000	3.92 ± 0.13
8k		506	530	66000	2.02 ± 0.04
8l		525	553	39000	3.54 ± 0.07
8m		526	557	48000	3.09 ± 0.04

^aAbsorption maxima, fluorescence emission maxima (both in nm), and molar absorptivity were measured after saponification of 8a–m and dilution into 150 mM KCl buffered with 10 mM HEPES and supplemented with 50 μM EDTA at pH = 7.22.

^bIn cell on/off ratios were determined using HEK-293 cells co-expressing GIRK1 and GIRK2 treated with 8a - m for 1 h and then subjected to Tl⁺ stimulus. Values represent the maximum fold increase in fluorescence after Tl⁺ addition (n = 6).

To determine the in vitro photochemical properties of the active Tl⁺ sensing species, 8a – m were treated with base to remove the acetoxymethyl esters and O-aryl acetate and then diluted into 10 mM HEPES buffered 150 mM KCl with 50 μM EDTA at pH = 7.22. Saponified 8a – m exhibited photochemical properties comparable to previously reported rhodols.^27,28 (Table 1).

The performance of rhodol Tl⁺ sensors in the Tl⁺ flux assay was assessed by incubating HEK-293 cells co-expressing G-protein-gated inwardly rectifying K⁺ (GIRK) channel subunits GIRK1 and GIRK2 with 8a – m in dye loading buffer for 1 hour prior to addition of 1 mM (final concentration) of Tl⁺ stimulus and the fold increase in brightness calculated. Dimethyl rhodol 8c demonstrated a 4.21-fold increase in brightness in the presence of Tl⁺ in HEK-293 cells. A notable decrease in ability to sense Tl⁺ among the dialkyl series 8c – 8f as the alkyl chain length increases was observed. Monoalkyl rhodols 8g and 8h as well as cyclic amine containing rhodols 8j, 8l, and 8m exhibit >3 fold increases in brightness upon addition of Tl⁺ stimulus. Gemdifluoro 8k exhibited almost half the response to Tl⁺ compared with the non-fluorinated analog 8j. Phenyl rhodol 8i exhibited very low initial fluorescence in HEK-293 cells requiring a longer exposure time than the other dyes and a minor increase in fluorescence upon addition of Tl⁺. Fluorescence was not detected in HEK-293 cells incubated with acetamide 8b before or after the addition of Tl⁺. Considering photochemical properties and response to Tl⁺ in HEK-293 cells, dimethyl rhodol 8c was identified as the best of the series and further evaluated for its utility as a reagent for Tl⁺ flux assays.

Cytoplasmic localization of Tl⁺ indicators is desirable for Tl⁺ flux assays of ion channels and transporters localized to the plasma membrane. The cellular localization of 8c was compared to Thallos by treating GIRK1 and GIRK2 co-expressing HEK-293 cells with Thallos, 8c, or the mitochondrial stain Rhodamine 123 for 1 h and all counterstained with the nuclear stain Hoechst. 8c exhibited diffuse cytoplasmic staining comparable to Thallos, in contrast to the punctate mitochondrial staining observed with Rhodamine 123 (Figure 2).

Figure 2.

Cellular localization of Thallos (a), 8c (b), and Rhodamine 123 (c). Confocal microscopy images were obtained in HEK-293 cells co-expressing GIRK1 and GIRK2 following incubation with Thallos (1 μM), 8c(5 μM), or Rhodamine-123 (1 μM) for 1 h and counterstained with Hoechst 33342 (1 μg/mL). Scale bar = 10 μm.

The pK_as of 8c and Thallos were determined to be 5.23 and 6.52, respectively (Supplemental Figure 1) indicating that 8c may be superior to Thallos when conducting assays at the lower range of physiological pH.

The performance of 8c was evaluated in a Tl⁺ flux assay to generate a concentration-response curve (CRC) of the GIRK activator ML297.²⁹ HEK-293 cells co-expressing GIRK1 and GIRK2 were loaded with 5 μM of 8c for 1 h, treated with 0.01 – 10 μM of ML297, and then subjected to Tl⁺ stimulus 8c exhibited a 3.09 ± 0.004-fold increase in fluorescence at the highest ML297 concentration (Figure 3a). Normalized, control subtracted fluorescence intensity values were sampled along the curves following Tl⁺ addition at each concentration and a CRC was fit using GraphPad Prism (Figure 3b). The EC₅₀ was determined to be 257 ± 3.7 nM while the same procedure utilizing Thallos produced an EC₅₀ of 272 ± 23 nM, which was not statistically significantly different.

Figure 3.

Concentration response data of ML297 in HEK −293 cells co-expressing GIRK1 and GIRK2 obtained with 8c. (a) Fluorescence intensity data from each well were normalized to the averaged fluorescence intensity of the first 6 time points of the experiment and the normalized fluorescence data for each replicate time point were averaged (n = 12). (b) Concentration response curve derived from data in (a) by subtracting the vehicle control from each concentration series and sampling a time point several seconds after addition of Tl⁺ stimulus.

Using the same cell line and GIRK activator, 8c was also compared to the commercially available, red-shifted Tl⁺ sensor FluxOR Red (Molecular Probes). Following the manufacturer’s procedure, which requires supplementation of assay and stimulus buffers with a proprietary background suppressor, FluxOR Red exhibited a 5.37 ± 0.001-fold increase in fluorescence at the highest concentration of ML297 compared to Thallos (4.30 ± 0.008-fold increase) and 8c (3.09 ± 0.004-fold increase). However, omission of the FluxOR Red background suppressor resulted in a significantly diminished increase in fluorescence of 1.17 ± 0.007 at the highest ML297 concentration for FluxOR Red while Thallos and 8c only showed modest decreases in the absence of background suppressor (2.80 ± 0.009 and 2.87 ± 0.004-fold increases in fluorescence, respectively, Supplemental Figure 2). In addition, 8c exhibited a comparable dynamic range and EC₅₀ results using 544/24 nm excitation and 593/40 nm emission filters indicating 8c can also be used very effectively with longer wavelength, off-peak filters (Supplemental Figure 3).

Strict adherence to the FluxOR Red protocol may not be compatible with or convenient for assaying certain channels or transporters. For example, many buffers for Tl⁺ flux assays, including those in most commercial kits, utilize gluconate as the primary anion due to the insolubility of Tl⁺ in Cl⁻ containing buffers. However, gluconate-based buffers may not be optimal for all targets and assay systems.³⁰ In addition, the components of extracellular fluorescence masking dyes may affect the pharmacology of targets of interest. Tolerance of the Tl⁺ sensing dye towards varied assay conditions would allow for greater flexibility. The performance of 8c, Thallos, and FluxOR Red were compared using a bicarbonate-based stimulus buffer with 500 μM of Tl⁺ under two conditions. First, an automation compatible “no-wash” procedure was tested where the wash step after dye loading was omitted and all buffers were supplemented with 1 mM of Allura Red AC to suppress extracellular fluorescence. Under these conditions, 8c and Thallos exhibited 3.40 ± 0.009 and 3.83 ± 0.007-fold increases in fluorescence and ML297 EC₅₀s of 204 ± 5.2 nM and 219 ± 23 nM, respectively, were calculated. In contrast, FluxOR Red exhibited no change in fluorescence upon Tl⁺ addition and the data could not be fit to a CRC. Second, using similar conditions omitting Allura Red AC while employing a wash step after dye loading produced comparable results (Supplemental Figure 2, Supplemental Table 1). These data demonstrate the utility of 8c in assays where extracellular fluorescence masking dyes are not compatible or where “no wash” procedures are desirable.

Conclusions

In conclusion, dimethyl rhodol 8c has been identified from a series of rhodol-based Tl⁺ sensors as a versatile reagent for Tl⁺ flux assays. It localizes to the cytoplasm similarly to Thallos while exhibiting greater pH tolerance and excitation and emission spectra red-shifted ~30 nm. 8c exhibits an excellent dynamic range using both on-peak and longer off-peak filter sets and is compatible with a variety of assay conditions.