The yin and yang of intracellular delivery of amphipathic optical probes using n-butyl charge masking

Abstract

Monitoring and manipulation of ionized intracellular calcium concentrations within intact, living cells using optical probes with organic chromophores is a core method for cell physiology. Since all these probes have multiple negative charges, they must be smuggled through the plasma membrane in a transiently neutral form, with intracellular esterases used to deprotect the masked anions. Here we explore the ability of the synthetically easily accessible n-butyl ester protecting group to deliver amphipathic cargoes to the cytosol. We show that the size of the caging chromophore conditions the ability of intracellular probe delivery and esterase charge unmasking.

Life takes place in water, the “miracle solvent”, 110 times more polar than hexane. Such extreme polarity presents a solubility challenge for drugs, as these tend to be non-polar. The calcium (Ca²⁺) dye revolution, initiated in 1980 by Roger Tsien¹, introduced an even greater challenge for cellular delivery of small organic molecules². Ca²⁺-selective fluorophores such as quin2, etc. use four negative charges for Ca²⁺ complexation, and so are completely plasma membrane impermeable. Yet such probes were developed to measure intracellular [Ca²⁺]. Their delivery into the cytosol was effected by masking the crucial anionic charges with the acetoxymethyl (AM) ester protecting group³. Intracellular esterases, present in all cells, removed these protecting groups, trapping the Ca²⁺ dye in the appropriate intracellular space (the cytosol, Fig. 1). Leaving the plasma membrane intact is attractive as the cell is relatively unperturbed compared to other methods, such as microinjection or whole-cell patch-clamp. It is no exaggeration to say that the AM ester “trick” revolutionised the study of Ca²⁺ signalling in cells, with tens of thousands of reports using this simple procedure. Three years before Tsien, Engels and Schlaeger used a wide variety of phosphate esters to deliver pro-cAMP molecules into cells as alkyl esters, with intracellular esterases liberating the second messenger, causing mast cell degranulation⁴. In 1977, the Tsien group went on to make a detailed study of the hydrophobicity of the alkyl chain of the acetoxy-alkyl protecting group, finding that butyryloxymethyl was best for inositol polyphosphates when the 2-, 3- and 6-hydroxyls were unprotected⁵.

Fig. 1. AM ester intracellular probe loading.

Cartoon of AM ester cell loading: step 1 cell entry; step 2 sequential de-esterification; step 3 complete liberation of probe.

The core challenge is the relative solubility of the fully protected compound (“X-AM_n” in Fig. 1) in the plasma membrane lipid bilayer versus the intracellular aqueous environment. Additionally, one must be able to have a practical synthesis of “X-AM_n”probes. Furthermore, if AM esters are not used, then the alternative must be rapidly removed by intracellular esterases.

In 2016, we introduced a new photochemical protecting group using an extended π-electron nitro aromatic chromophore called “BIST” (for bisstyrylthiophene⁶). Our first application of BIST was to Ca²⁺ uncaging⁶. We appended two EGTA chelators to create a new caged Ca²⁺ probe (hence BIST-2EGTA, 1, Scheme 1), as we were concerned about the aqueous solubility of simpler molecules such as BIST-1EGTA (2). During the development of BIST-2EGTA, we attempted to make its AM ester, as loading caged Ca²⁺ probes into intact cells has proved a useful tool in optical physiology^7–10. For reasons we do not understand, such efforts were fruitless. Recent work by the Schnermann group¹¹ drew our attention to the idea to use n-butyl as a masking group of anionic charge. They showed one could deliver a caged compound with two latent negative charges (carboxylates) to the cell interior successfully when they were masked with n-butyl. A further attraction was that we could preload n-butyl onto various esters, something that is not feasible for the AM group, making the n-butyl group much more synthetically tractable (see below). Here, we describe the yin and yang of using n-butyl for caged IP₃ and caged Ca²⁺.

Scheme 1. Caged probe structures and synthesis.

(A) Structure of BIST-2EGTA; (B) Synthesis of BIST-1EGTA, reagents: (a) 2-(4-Nitrostyryl)-5-vinylthiophene, Pd(Ac)₂. (b) Ph₃P/NaOH. (c) Ethyl bromoacetate. (d) KOH. (e) n-Butyl iodoacetate; (C) Structure of various NV-IP₃ probes.

Our first test of n-butyl as a substitute for the AM ester was with 6-nitroveratryl(NV)-inositoltrisphosphate(IP₃), since the AM version of this probe is effective for intracellular delivery of caged IP₃ molecules^12–14. The hexakis-n-butyl protected NV-IP₃ compound (3a) was synthesised directly from the triol precursor¹⁵ in a 79% yield (Supplementary information). This can be compared to the 20% we reported¹² for the conversion of the triol into 3b. This improvement in yield of the of the cell permeable caged compound shows one potential benefit of using n-butyl in place of the AM protecting group. We tested the biological effectiveness of this probe by using the mobilisation of Ca²⁺ release from intracellular stores by two-photon photolysis (2PP) of 3a in astrocytes.

First, we established what power could be used before any phototoxic responses could be detected (Fig. 2A). We directed a 3×3 grid of light flashes at 720 nm from a Ti:sapphire laser at the cell body with a range of 10 to 45 mW. Powers around 40–45 mW generated Ca²⁺ signals, consistent with our previous in vivo study¹². So we used 10–20 mW for uncaging in astrocytes. Next, a low [3a], and a Ca²⁺ dye, were bath applied to brain slices acutely isolated from juvenile mice (Fig. 2B). Robust intracellular Ca²⁺ transients were induced in the irradiated cells (Fig. 2B). Similar results were seen when the bolus injection method was used to load 3a and a Ca²⁺ dye into slices isolated from young adult mice (Fig. S1). While it is impossible to make a quantitative comparison of the effectiveness of the NV-IP₃ AM (3b) and n-butyl (3a) probes, as the caged compounds are non-fluorescent and loading is inherently stochastic, our IP₃ uncaging data from photolysis of 3c derived from 3a inside cells (Figs. 2B, S1) were very encouraging. These are the first examples, to our knowledge, of using n-butyl to load a caged second messenger into living cells.

Fig. 2. 2P photolysis of IP₃ within intact astrocytes.

Astrocytes in brain slices were loaded with a fluorescent Ca²⁺-dye without (A) or with (B) 3a. Changes in [Ca²⁺] were monitored by 2P imaging after irradiation with 3×3 grid of 20 ms flashes of 720 nm light. (A) 2PP toxicity was established by irradiation of the cell (red arrow) with increasing power levels, as indicated. (B) 2PP of 3a in astrocytes evokes Ca²⁺ signals detected by fluorescence imaging.

We synthesised BIST-1EGTA, and its n-butyl ester, as shown in Scheme 1. Starting with the known⁶ azido-alkene 4, the thio subunit was added by a Heck reaction in 77% yield to give 5. This was deprotected with Ph₃P followed by NaOH, to give its analogous diamine, which was elaborated into BIST-1EGTA (as the K⁺ salt) 2 or its n-butyl ester 6, as shown in Scheme 1. It is important to note that 6 could be made directly in two steps (35% yield) from 5 using n-butyl iodoacetate as the alkylating agent in the last step (e) shown in Scheme 1. In contrast, the steps to produce DMNPE-4/AM from its diazide¹⁶ were effected in an isolated yield of only 4.5% (Ellis-Davies, unpublished).

The physicochemical properties of BIST-1EGTA were similar to BIST-2EGTA⁶ (Fig. 3 and Supplementary Information). BIST-1EGTA has an extinction coefficient of 51,000 M⁻¹ cm⁻¹ with a λ_max = 445 nm. The Ca²⁺ affinity of BIST-1EGTA at pH 7.5 was 8.9 nM. The quantum yield of photolysis was 0.31. Finally, 2PP at 810 nm (mode-locked, 120 fs pulses) revealed a quadratic dependence of Ca²⁺ release upon power (Fig. 3B). Importantly, when the Ti:sapphire laser was taken out of mode-lock, irradiation with near-IR light caused no uncaging (violet dots in Fig. 4B). These similarities are important, as it assured us that if we could deliver BIST-1EGTA into intact cells via its n-butyl ester (6), the de-esterified probe (2) would be a very effective caged Ca²⁺ for 2PP.

Fig. 3. Absorption and 2P photolysis of BIST-1EGTA (2).

(A) Absorption spectra of 2 (Hepes, pH 7.2) and 6 (DMSO). (B) Power dependence of Ca²⁺ release from 2 using a Ti:sapphire laser in (black dots, mean +/− S.D., n = 5) and out (violet dots) of mode-lock. The red line is the quadratic fit to the former, with dashes showing 95% confidence.

Fig. 4. Loading of BIST-1EGTA/Bu (6) into intact cardiac myocytes.

Panels (A) and (B) show confocal, line-scan images of myocytes filled with rhod-2. 2PP was at 810 nm for 10 ms. Changes in fluorescence are illustrated on a pseudo colour scale with warmer colours indicating higher [Ca²⁺]. (A) Ca²⁺ signals following electrical field stimulation (0–4 s) and 2PP (white box, see B). (B) Expansion of white box from (A) showing a spontaneous Ca²⁺ spark and the artefact from point 2PP. (C) Images of intact myocytes with (i, ii) and without (iii, iv) bath application of 40 μM 6. (i, iii) are green fluorescence and (ii, iv) bright field images.

We chose cardiac myocytes for this assay, as we have shown that 1 works well for 2PP in these cells⁶. The procedure for loading 6 into cardiac myocytes followed that used for 3a in astrocytes (Supporting information). The core idea is that even quite hydrophobic compounds are sparingly soluble in the physiological buffer surrounding cells, when partnered with the neutral detergent Pluronic-127. Compounds then partition into the plasma membrane, a process which allows a small amount of the fully protected probe to enter the cytosol, where intracellular esterases rapidly de-esterify it, generating a large concentration gradient to drive more fully protected probe into the cell (Fig. 1). This allows millimolar levels of Ca²⁺ dyes, for example, to be concentrated inside intact cells in a very efficient and simple way. Typically, this protocol is split into two 30–60 mins periods, one for loading and a second for complete de-esterification. The many 10,000s of Ca²⁺ imaging experiments reported over the past 40 years testify to power of this method. Its success is predicated on the esters being adequately exposed to intracellular esterases present in the cytosolic solution. When 6 was applied to cardiac myocytes at 20 μM, irradiation with 20 mW for as long as 10 ms produced no detectable signals from 2P-induced uncaging (Fig. 4A,B).

We varied the loading conditions considerably to improve this disappointing result. For example, we increased the loading time to as much as 180 mins, added BSA to the external solution, used high temperatures (37°C), and tried HeLa cells in place of cardiac myocytes. In all cases we had as the control the loading of a fluorescent Ca²⁺ dye as its AM ester. In every case electrical depolarisation induced large intracellular Ca²⁺ signals (Fig. 4A), implying 6 is non-toxic, however we did not detect robust signals from 2PP on any occasion. Only small artefacts from 2P excitation of the Ca²⁺ dye were seen (Fig. 4B). We were both surprised and disappointed by these results. We know that BIST is non-fluorescent in water, but is modestly fluorescent in organic solvents such as DMSO^{6, 17}. Thus, we imaged cells treated with 6, but before Ca²⁺ dye loading, with the blue laser on our confocal microscope. We detected very bright staining of all membranes of the cardiac myocytes (Fig. 4C). We infer from these data that 6 is so lipophilic that it does not allow exposure of sufficient material to intracellular esterases for effective deprotection. Thus, even though we had made a BIST analogue of 3a (data not shown), considering the results shown in Fig. 4, we decided not to perform any cellular tests of this molecule.

We did confirm that when the highly water soluble form of BIST-1EGTA, namely compound 2, was applied to myocytes, it could be used for very effective 2PP in the intracellular space to produce Ca²⁺-induced Ca²⁺-release (CICR, the fundamental signalling mechanism inside heart muscle cells that causes contraction). To do this we followed our recently developed approach¹⁸ of using bath application of a low concentration (50 μM) of BIST-caged Ca²⁺ to load cardiac myocytes permeabilised with β-escin (Fig. 5A). Similar to BIST-2EGTA, we could induce large CICR signals with low power (20 mW) and short uncaging periods (1 ms, Fig. 5B). These conditions must be compared to those we had to use with our NV-based caged Ca²⁺, DM-nitrophen, namely 50–80 mW for 50 ms^{19, 20}. The significant reduction in 2P power dosage (800-fold) is a result of the large 2P cross-section of the BIST chromophore. Note that historically we used 1–2 mM DM-nitrophen for 2PP^{19, 21}, whereas here we used only 0.05 mM BIST-1EGTA. While traditional photochemical protecting groups such as NV used in DM-nitrophen show excellent 1PP efficiency^{22, 23}, the relatively poor 2P cross-section of NV required a comparatively high concentration for 2PP. In contrast the excellent 2P cross-section of BIST^{6, 17} enables us to use quite low concentrations of the caged compound 2, yet still release large amounts of Ca²⁺ using 2PP. We recently reported similar improvements in efficacy with our BIST-caged GABA¹⁷.

Fig. 5. 2P photolysis of BIST-1EGTA (2) in permeabilised cardiac myocytes.

2 was applied to a permeabilised myocytes at 0.05 mM with rhod-2 as their anions solubilised in physiological buffer at pH 7.2. 2PP was at 810 nm for 1 ms. Changes in fluorescence are illustrated on a pseudo colour scale with warmer colours indicating higher [Ca²⁺]. (A) Fluorescence (left) and bright field (right) images of a myocyte with the line-scan (yellow) and uncaging positions (red) shown. (B) Confocal line-scan image time-series with the spatial spread in x (upper), and the intensity (taken from the black boxed segment) shown in the lower panel. Note the 2P-initiated Ca²⁺ release from the internal Ca²⁺ store was much larger than a Ca²⁺ spark seen in Fig. 4B.

In conclusion, it seems the equilibrium between the lipophilicity of amphipathic optical probes such as the tetrabutyl ester of BIST-1EGTA and the ability of cells to de-esterify such molecules lies very much to the left. In contrast, the n-butyl protecting group can be used with more traditional caged compounds, such as NV-IP₃, quite effectively. So while the exceptional 2P cross-section of BIST⁶ allows very mild irradiation conditions with living cells (Fig. 5 and refs^6,18), the inherent lipophilicity of the large, extended π-electron chromophore might constrain its application to highly water soluble probes such as G3.5-BIST-GABA¹⁷ and BIST-caged calciums such as 1 and 2 in future biological applications. Importantly, the highly effective cell stimulation of Ca²⁺ release using 2-photon photolysis of BIST-1EGTA implies this new probe could be useful for many physiological experiments beyond cardiac muscle, as photorelease of Ca²⁺ in living cells has been used in, for example, neuronal, glia, and pancreatic cells ²³.