Synthesis of Fluorescent Jasplakinolide Analogues for Live-Cell STEDa Microscopy of Actin

The nanometer thickness of filaments and the dynamic behavior of actin—a protein playing a crucial role in cellular function and motility—make it attractive for observation with super-resolution optical microscopy. We developed the solution-phase synthesis of des-bromo-des-methyl-jasplakinolide-lysine, used as the “recognition unit” (ligand) for F-actin in living cells. The first amino acid—Fmoc-O-TIPS-β-tyrosine—was prepared in 78% yield (two steps in one pot). The new solution-phase synthesis involves 2-phenylisopropyl protection of the carboxyl group and does not require excesses of commercially unavailable amino acids. The overall yield of the target intermediate obtained in nine steps is about 8%. The 2-phenylisopropyl group can be cleaved from carboxyl with 2–3% (v/v) of TFA in acetonitrile (0–10 °C), without affecting TIPS protection of the phenolic hydroxyl in β-tyrosine and N-Boc protection in lysine. Des-bromo-des-methyl-jasplakinolide-lysine was coupled with red-emitting fluorescent dyes 580CP and 610CP (via 6-aminohexanoate linker). Actin in living cells was labeled with 580CP and 610CP probes, and the optical resolution measured as full width at half-maximum of line profiles across actin fibers was found to be 300–400 nm and 100 nm under confocal and STED conditions, respectively. The solution-phase synthesis of des-bromo-des-methyl-jasplakinolide-lysine opens a way to better fluorescent probe perspective for actin imaging.


■ INTRODUCTION
Actin protein plays a crucial role in cellular function and motility. 1 It can be present either as a monomer (G-actin; globular) or, upon polymerization, it may form filaments (Factin): flexible fibers with a diameter of 4−7 nm and length of up to several micrometers. In living cells, both forms of actin are present in equilibrium; they are essential for the proper mobility and contraction of cells during cell division, cell motility, cytokinesis, vesicle and organelle movement, cell signaling, as well as the establishment and maintenance of cell junctions and cell shapes. The nanometer thickness and dynamic behavior of actin filaments make them an attractive object for observation with super-resolution optical microscopy.
The fluorescent probes for super-resolution and live imaging of actin 2−5 incorporate the so-called des-bromo-des-methyljasplakinolide-lysine (Figure 1), 6 as the ligand or "recognition unit" for F-actin in living cells. This macrocyclic depsipeptide has a reactive amino group, and its salts can be readily generated from N-tert-butoxycarbonyl derivative  which represents the key intermediate and stable precursor of the conjugates with organic dyes. Compound 7-H is commercially unavailable, and the solid-phase synthesis of 7-H has been outlined only briefly. 2,6 The aim of the present work was to develop the new and productive route to macrocyclic depsipeptide 7-H, compare the syntheses on a solid phase and in solution, prepare the conjugates of compound 7-H with fluorescent dyes, and apply them as fluorescent probes for the super-resolution microscopy of actin filaments in living cells. As cell-permeate fluorescent dyes, we have chosen carbopyronines 580CP and 610CP which demonstrated high imaging performance as conjugates with various ligands. 5,7,8 The absorption and emission spectra of these dyes are given in Figure 1 and the photophysical properties in Table 1.

■ RESULTS AND DISCUSSION
Both synthesis routeson the solid phase and in solution involve Fmoc-O-TIPS-β-tyrosine as the first amino acid (AA1; Schemes 1 and 2). This compound was initially obtained via a multistep procedure including the Michael addition of a chiral dibenzyl amine to p-coumaric acid ester, separation of the diastereomers, N-debenzylation, and manipulation with Oprotecting groups. 6 We found a shorter route to Fmoc-O-TIPS-β-tyrosine, which starts from commercially available Fmoc-β-tyrosine (Scheme 2). The two-step procedure includes silylation with triisopropylsilyl chloride on both oxygen centers 9 followed by the hydrolytic cleavage of the more labile (triisopropylsilyl)ester group under mild basic conditions and affords the required amino acid AA1 (Scheme 2). The solidphase synthesis (Scheme 1) provides triamide 1-H-TIPS as the key intermediate. Compound 1-H-TIPS was isolated with an overall yield of about 22% (52%, when calculated on the loading degree of the first amino acid AA1). However, these yields are based on the use of large excess of amino acids AA1, AA2, and AA3. The first two are not commercially available and have to be prepared separately. Therefore, the use of large excess of AA1 and AA2 is not cost-and time-efficient. Another important detail of the solid-phase synthesis is that the cleavage from the resin is performed in the presence of weakly acidic hexafluoroisopropanol. The latter (b. p. 58°C) concentrates in the reaction mixture in the course of solvent evaporation (DCM) and causes (partial) removal of the triisopropyl silyl group. We added a higher boiling solvent (ethyl acetate) into the solution in order to prevent this undesirable effect and suppress the formation of deprotected phenol 1-H-H.
Planning the solution-phase synthesis of compound 1-H-TIPS (5-H-H in Schemes 2 and 3), we realized that its success is determined by the correct choice of carboxyl protection in AA1 (Scheme 2). We used the 2-phenylisopropyl protecting group 10 because 2-chlorotrityl esters 11 partially cleaved in the course of work-up and isolation procedures (chromatography), when the synthesis was carried out according to Scheme 2. The synthesis was carried out according to Scheme 2, using watersoluble carbodiimide (free base) in the presence of HOAt and 2,4,6-collidine in DCM. 12 Under these conditions, no racemization was observed. 12 9-Fluorenylmethylcarbamate groups were cleaved using diethylamine (the excess of which was removed by several evaporations with toluene); intermediate compounds with free amino groups were not isolated but used directly in the following amidation reactions. In this approach, the excess of N-protected amino acids is not required (which is an advantage over the solid-phase methodology). The final stepremoval of 2-phenylpropyl protecting groupwas effected using 2−3% solution of TFA in acetonitrile at 0... +5°C. Under these conditions, N-tertbutoxycarbonyl protection of amines is stable. 10,13 Moreover, TIPS protection of the phenolic hydroxyl group turned out to be stable as well. However, we detected and isolated compound 5-H-C(CH 3 ) 2 C 6 H 5 (17%), which was formed when 2-phenylpropyl residue was transferred to another nucleophilic centernitrogen atom of tryptophan. The synthesis in solution is attractive not only because of the relatively high overall yield (35%; Scheme 2) but also because it is not necessary to apply (unrecoverable) excess of exotic and expensive amino and (S)-2,4-dimethylpent-4-enoic acids. The final steps of the assembly of macrocyclic depsipeptide 7-H are common for the solid-and solution-phase syntheses and are given in Scheme 3. The first reactionformation of ester 6 from carboxylic acid 5-H-H (1-H-TIPS in Scheme 1) and (S)-5-hexen-2-ol in the presence of carbodiimide (EDC*HCl) requires 4-(N,N-dimethylamino)pyridine (DMAP) as a catalyst. We found that the use of more than 10 mol % of Figure 1. Des-bromo-des-methyl-jasplakinolide-lysine for conjugation with cell-permeate fluorescent dyes; absorption and emission spectra of 580CP and 610CP are shown (see also  4.78 ppm) in the spectrum of 7-H. However, we do not have a plausible explanation of this 1,2-shift. 5-Hexen-2-ol did not contain appreciable amounts of 2-methyl-5-hexen-1-ol. The overall yield of compound 7-H obtained in nine steps according to Schemes 2 and 3 is about 8%. Amine 9 (Scheme 4) was prepared from compound 7-H in the presence of formic acid. Deprotection with formic acid was found to be cleaner than the cleavage of the tertbutoxycarbonyl group with trifluoroacetic acid (TFA).
Conjugates of fluorescent dyes 580CP and 610CP (their spectra are given in Figure 1) with ω-aminocaproic acid (linker) and amine 9 (actin ligands) were obtained in three steps, as outlined in Scheme 4 (via N-hydroxysuccinimidyl esters; 3,4 for details, see the Supporting Information and ref 5). We labeled actin in living human osteosarcoma cells (U-2 OS) (Figure 2A,B) and in kidney cells derived from the African green monkey (COS-7) ( Figure 2C,D) using 580CPjasplakinolide ( Figure 2A,C) and 610CP-jasplakinolide ( Figure  2B,D) probes, respectively. Both probes (for structures, see Scheme 4) performed well in confocal and STED (stimulated emission depletion) microscopy. The optical resolution (full width at half maximum of a line profile) in the STED mode improved: the apparent diameters of actin fiber bundles under confocal and STED conditions were 300−400 nm and ca. 100 nm, respectively (see Figure 2). Both dyes (for spectral properties, see Table 1) have some residual emission at 775 nm (wavelength of the STED laser; see Figure 1) but virtually no absorption at this wavelength. These valuable spectral features provide an efficient STED effect and, as a result, optical resolution improvement without undesirable reexcitation with the STED beam. Importantly, the conjugates of carbopyronine dye 580CP enable two-color STED microscopy in living cells with standard optical settings (e.g., in combination with SiR dye; see Table 1). 5,7 ■ CONCLUSIONS In vitro labeling of actin filaments ( Figure 2) with 580CP-and 610CP-jasplakinolide conjugates exhibits different patterns in different cell lines and at different concentrations: the best imaging results were achieved when 580CP probe was applied at 5 μM for 30 min and 610CP-jasplakinolide−at 1 μM for 60 min. Compared to 610CP-, 580CP-jasplakinolide enables an enhanced labeling of intricate actin structures. The overall performance can be affected by the specific dye residue coupled to the (same) jasplakinolide ligand, as the dye was shown to influence the core characteristics of the whole fluorescent probe, such as binding parameters (kinetics, affinity, equilibrium between F-and G-actin), cytotoxicity, and, most importantly, cell entry and/or retention. 5 Other (less toxic, more specific, brighter) fluorescent probes for actin in cells and tissues may help further to understand the role of this protein in cell functions and motility. 16 The proposed methodology enables the scalable synthesis of compounds 7-H, 9, their analogs (e.g., by varying the structure of unsaturated alcohol in Scheme 3), and their conjugates with fluorescent dyes, in order to reveal new important aspects of actin behavior in the living matter.
Fmoc Cleavage. The solvent was removed and the resin dried for 12 h at 0.2 mbar. Two portions of the resin (5.5 mg each) were used to determine the loading degree of AA1 (0.44 mmol g −1 ) using a mixture of DBU/piperidine/NMP (2:2:96 v/v) for cleaving the Fmoc group and measuring the optical density (at 304 nm) against the blank sample. 15 The main part of resin (1.25 g, 0.55 mmol) was washed with NMP (9.4 mL) and subjected to deprotection by refilling the syringe with a mixture of DBU/piperidine/NMP (2:2:96 v/v) and shaking (300 rpm) for 10 min at 23°C. This operation was repeated (with 30 min exposure at 23°C). Completion of the cleavage was controlled by TLC (hexane/EtOAc 75:25): the application of the second cleavage cocktail revealed no appreciable UV active spot(s) of the Fmoc derivatives.
Activation and Coupling. AA2 (AA3, A4) (1.65 mmol, 3 equiv) was added into an oven-dried round bottom flask filled with argon and dissolved in NMP (7.5 mL). Then, a solution of HOBt (4.0 equiv, 2.2 mmol, 299 mg) in NMP (750 μL) was added followed by DIC (4.0 equiv, 2.2 mmol, 343 μL), and the reaction solution was stirred for 5 min under argon. The syringe with the resin was filled with the reaction solution; NMP (1 mL) was used for rinsing the round-bottom flask. The suspension was shaken (300 rpm) for 20 h at 23°C. The syringe was drained, the resin was washed with NMP (2 × 9.4 mL), CH 2 Cl 2 (3 × 9.4 mL), and subjected to deprotection.