Synthesis of Thioxanthone 10,10-Dioxides and Sulfone-Fluoresceins via Pd-Catalyzed Sulfonylative Homocoupling

Our report describes the facile and scalable preparation of 9H-thioxanthen-9-one 10,10-dioxides via Pd-catalyzed sulfonylative homocoupling of the appropriately substituted benzophenones. This transformation provides a straightforward route to previously unreported sulfone-fluoresceins and -fluorones. Several examples of these red fluorescent dyes have been prepared, characterized, and evaluated as live-cell permeant labels compatible with super-resolution fluorescence microscopy with 775 nm stimulated emission depletion.

S ulfone-fluorescein (Scheme 1a; X = SO 2 ) is the sulfone- bridged analogue of the long-established green-emitting fluorescent dye fluorescein (3′,6′-dihydroxyfluoran).The introduction of a strongly electron-deficient bridging group has been previously demonstrated to shift the absorption and emission maxima, mainly by decreasing the LUMO energy level of the fluorophore, 1a into the red range (>600 nm) preferred for a greater light penetration depth and lower phototoxicity.Unlike the spirolactone variants with a modified xanthene core (thiofluorescein, 2a carbofluorescein, 2b Si-fluorescein, 2c and their halogenated versions 2d ), only the derivatives with a 2alkyl-or 2-alkoxyphenyl pendant ring have been reported for bora-fluorescein 2e and phospha-fluorescein, 1 while sulfonebridged fluorone 3 analogues of fluorescein remain unknown outside of the patent literature. 4 concise synthetic strategy for accessing sulfone-fluoresceins would involve nucleophilic addition of aryllithium or arylmagnesium reagents to the keto group of appropriately substituted 9H-thioxanthen-9-one 10,10-dioxides.These heterocycles have found use in organic electronics as acceptor units 5 in building blocks for thermally activated delayed fluorescence emitters, in particular in challenging recent applications such as phosphorescent 6 and circularly polarized OLEDs. 7hioxanthone 10,10-dioxides have previously been prepared through nucleophilic 8 or electrophilic 9 ring closure of diaryl sulfones obtained in a multistep sequence or, by far most commonly, via oxidation of preassembled thioxanthones, 7,10 thus limiting the scope of compatible functional groups.Instead, we envisaged an alternative synthetic approach involving a Pdcatalyzed sulfonylative homocoupling of benzophenones bearing the leaving groups at positions 2 and 2′, readily accessible via established methods 11 (Scheme 1b).For this, we were inspired by the recent reports of Wu 12a and Wang and Jiang 12b relying on sodium dithionite as a masked "SO 2 2− " synthone in their preparation of alkyl aryl sulfones, because the reported systems employing other SO 2 surrogates 13 (K 2 S 2 O 5 , DABSO, or formamidinesulfinic acid) consistently failed when tested with our substrates.
Gratifyingly, after a brief investigation of the reaction conditions (Table S1), we determined that the target thioxanthone 10,10-dioxides formed in good yields [>70% for many examples (see Scheme 2)] with an air-stable and inexpensive Pd(II) catalyst Pd(dppf)Cl 2 when DMSO was used as the reaction solvent with mild heating (80 °C).Distinct from the reported conditions, 12a no addition of an external base or quaternary ammonium salts was needed.At least 1.5 equiv of Na 2 S 2 O 4 was required to achieve complete conversion of the starting material, but a larger excess of dithionite was well tolerated and found necessary in certain cases.Decreasing the Pd catalyst loading resulted in substantially lower reaction yields.
The reaction was sensitive to the nature of the solvent, and even though other dipolar aprotic solvents (in particular DMF) were suitable, the reproducibility suffered because of the significant induction time as determined by means of in situ IR spectroscopy (see Figure S1).The substrates bearing strong electron-withdrawing groups yielded xanthones (e.g., 2m′ and 2o′) as the major products, likely via competing hydrolysis of aryl triflates to phenols if the main reaction was slowed, and the presence of aryl halides (as in 1r) other than fluorine was not tolerated.Most peculiarly, and in stark contrast with previous reports, the transformation was successful only in the intramolecular version; otherwise, alternative reactivity with the formation of mixtures containing diaryl sulfides (2p) or disulfides (2q) was noticed, while other bridged (1t and 1u) or simple aryl triflates (1v) were unreactive.
The required presence of a coordinating 2-carbonyl group suggested the initial formation of an oxidative addition complex (i) [exemplified by S11 (Scheme 3)], possibly stabilized via chelation, which should then undergo an insertion of SO 2 2− (sulfoxylate dianion arising by disproportionation from dithionite dissociated into two sulfur dioxide radical anions 14 ) generating the corresponding Pd(0)-chelated arylsulfinate intermediate (ii).In the absence of an intramolecular electrophile (i.e., in the case of monotriflate 1u), the intermediate complex dissociates to form the free sulfinate product (iii).This product was trapped by alkylation with an excess of CH 3 I, leading to aryl methyl sulfone 2u in 58% yield.Otherwise, the second (intramolecular) oxidative addition of an aryl triflate (for 1a) or bromide 15 (for 1a′) leads to the intermediate (iv), which then undergoes reductive elimination of 2a, regenerating the active catalyst.In a control experiment with 2-bromobenzophenone, it was confirmed that only aryl triflates underwent the initial oxidative addition under these reaction conditions, as no formation of 2u was detected upon addition of excess CH 3 I electrophile.
The robustness of the proposed method allowed us to prepare 3 (Scheme 4), the key building block for the synthesis of sulfonefluoresceins, on a multigram scale by demethylation of 2h in a three-step sequence starting from the commercial ultraviolet absorber benzophenone-6 zophenone), in high yield and purity without chromatographic separation.While several protecting groups 2b, 16 were evaluated for acidic phenol groups of 3, TBS-protected compound 4a was the substrate of choice for the preparation of unsubstituted sulfone-fluoresceins 5a and 5b and sulfone-fluorone 5c (Scheme 4a).The photophysical properties of these fluorophores are compiled in Table 1 and Figures S2 and S3.
Both dyes 5a and 5b have absorption maxima within the range of 620−630 nm (nearly optimal for excitation with a 630−640 nm pulsed laser) and are characterized by far-red emission; however, their quantum yields in aqueous buffer are modest.The fluorescent forms of sulfone-fluoresceins and sulfonefluorones undergo protonation closing into the colorless spirolactones (for 5a, with a pK a of ∼8) or form triarylmethanol water adducts (for 5b and 5c, with corresponding pK a values of ∼5−6); in particular, the color of the anionic form of 5c quickly fades in solution.This electrophilic reactivity of the sulfonefluorone core also accounts for lower than expected values of molar attenuation coefficients for sulfone-fluoresceins (ε = 3 × 10 3 M −1 cm −1 at pH 9 for 5a vs ε = 9 × 10 4 M −1 cm −1 at pH 10 for fluorescein 17 ).
For the preparation of the fluorophores tagged with a free carboxylic acid (suitable for conversion into targeted labels for fixed or live-cell imaging via attachment to a suitable highaffinity ligand), di-O-benzyl-protected 4e was chosen for the technical ease of separation and maintained orthogonality to the tert-butyl ester group.As a test ligand, the bioorthogonal ωchloroalkane/HaloTag protein system 18 was selected for its extremely high reaction rates with triarylmethane dye-derived probes (k app values of 1.0 × 10 6 M −1 s −1 for fluorescein ligand 19a and approaching 10 8 M −1 s −1 for certain rhodamines 19b ), leading to rapid and irreversible covalent linking within the livecell environment.Benzyl-protected sulfone-fluorescein carboxylic acids 6a−6c were coupled to the HaloTag(O2) amine, and target fluorescent probes 7a-Halo−7c-Halo were liberated by hydrogenolysis followed by sequential treatment with TFA and DDQ (Scheme 4b and Figure S4).
Test labeling was performed in living U2OS-Vim-Halo cells engineered using the CRISPR-Cas technology, which were preferred for their high cell-to-cell reproducibility as opposed to transient transfection methods. 20HaloTag-fused vimentin (a cytoskeletal structural protein) was stained with 7a-Halo (500 nM overnight in complete cell growth medium), followed by two washing steps (30 min each).Cells were imaged live (Figure 1) or after fixation with paraformaldehyde (Figure S5).Relatively long integration times [line or frame accumulation (see Table S2  compatibility of the label with stimulated emission depletion (STED) super-resolution fluorescence imaging.The attainable resolution was limited by the low molecular brightness of fluorophore 5a and the dynamics of intermediate filaments in living cells, despite the distinct fluorogenic response of the dye upon binding to HaloTag7 protein (Figure S6).Therefore, livecell labeling with postfixation, permitting a longer total imaging time on immobile structures, was attempted and indeed demonstrated improved resolution (Figure S5).The samples stained with 7b-Halo and 7c-Halo were characterized by a significantly inferior quality of labeling, making the imaging impracticable.
In summary, we have developed an original entry in the synthesis of sulfone-fluorescein fluorophores and performed their photophysical characterization and preliminary evaluation as live-cell compatible far-red-emitting fluorescent labels.The proposed synthetic approach to thioxanthone 10,10-dioxides increases the availability of these building blocks for organophotocatalysis, as well as for photovoltaics, electroluminescence, and other material science applications.Furthermore, we are currently exploring variations of the reported reactivity toward the synthesis of other sulfone-embedded heterocycles.
Scheme 1.(a) Fluorescein Analogues with a Xanthene Ring Modified by Varying Bridging Groups and (b) Modular Synthetic Approach to Thioxanthone 10,10-Dioxides a

Figure 1 .
Figure 1.(A) Two-color overview confocal image of a living U2OS-Vim-Halo cell, labeled with probe 7a-Halo (500 nM, overnight; yellow) and nuclear stain Hoechst 33342 (3.6 μM, 10 min; cyan).(B) Confocal and (C) STED images of labeled vimentin filaments of the same cell.(D) Line profiles (average of five pixels) across a vimentin filament indicated with blue arrows in panel C, with the corresponding fits to a Gaussian function (for the confocal image) and a Lorentzian function (for STED).The corresponding full widths at half-maximum (fwhm) are indicated, demonstrating the resolution below the diffraction limit in panel C.The scale bars are 10 μm in panel A and 1 μm in panels B and C.

b
From 1a′. c At 100 °C for 8 h.d At 80 °C for 24 h.
a Optical properties measured in 0.1 M phosphate buffer (pH 9.0).b Fluorescence quantum yield.c Fluorescence lifetime.