Degradable linear and bottlebrush thioester-functional copolymers through atom transfer radical ring-opening copolymerization of a thionolactone

. The radical thiocarbonyl addition (TARO) with vinyl comonomers affords selectively degradable thioester functional polymers promising for biomedical applications. Herein, the use of atom transfer radical polymerization (ATRP) is investigated for the first time, using dibenzo[c,e]oxepane-5(7H)-thione (DOT), Cu(I)Br, and as and respectively, with the acrylate comonomers poly(ethylene glycol) methyl ether acrylate (PEGA), methyl acrylate, benzyl acrylate, and butyl acrylate. Polymerizations were impeded by a side reaction, the Cu(I)-catalyzed dethionation of DOT to its (oxo)lactone analog, which caused the loss of up to 50 mol% of DOT in the early polymerization stages and limited the final copolymer DOT content. Nonetheless, readily degradable copolymers with low dispersities (1.10 ≤ Ð ≤ 1.26) were formed using DMSO, acetonitrile, or toluene as solvent. Presuming adventitious water to be the oxygen source, the dethionation side reaction could be minimized ( ≥ 5 mol -% lactone) by using anhydrous polymerization conditions, which enabled the synthesis of copolymers with higher DOT content. Exploiting documented advantages of ATRP over thermally-initiated RAFT polymerization in the synthesis of brushes, water-soluble molecular brushes were prepared by grafting PEGA – DOT copolymers from a pre-made multi-ATRP initiator. Due to faster incorporation of DOT, the cleavable thioesters were located close to the junctions and enabled the fast (< 1 min) oxidative cleavage of the arms from the core to give water-soluble products using 10 mM oxone. Expanding the scope of the ATRP and TARO methods, this work presents facile access to polymer materials with tailored architectures and degradability. Scalable Synthesis of Single-Chain Nanoparticles under Mild Conditions.


Introduction
The radical ring-opening polymerization (RROP) of suitable cyclic monomers provides access to (co-)polymers featuring heteroatom backbone functionality including cleavable groups. 1,2 RROP unfolds its full potential in combination with the architectural control offered by reversible deactivation radical polymerization (RDRP) methods. Especially the combination of cyclic ketene acetals 3 with RAFT polymerization 4 has been exploited to prepare materials with high functional group tolerance, stimulus responsiveness, tailored architectures, and predictable backbone degradability. 5 However, cyclic ketene acetals have disadvantages. They show slow copolymerization behavior due to the high energy of the intermediate acetal carbon-based radical and usually need to be fed in excess to force incorporation during copolymerizations of 'moreactivated' (meth)acrylic monomers. Furthermore, the resulting backbone ester groups require harsh, non-selective degradation conditions.
Recently, thionolactones were shown to undergo radical ring-opening via thiocarbonyl additionring-opening (TARO) [6][7][8][9][10][11][12][13][14] which provides backbone thioester functionality. These thioesters can be selectively cleaved through aminolysis, thiolysis (including under biologically relevant conditions) 7 and, very rapidly, through persulfate oxidation. 6 RDRP of thionolactones has been achieved exclusively through RAFT polymerization with which TARO radical polymerization appears to be fully compatible. During the writing of this manuscript, Lages et al reported the successful nitroxide-mediated copolymerization of n-butyl acrylate and styrene with up to 2 mol-% DOT. 15 Atom transfer radical polymerization 16 is, arguably, more complex than RAFT polymerization. 17 But it has a distinct advantage over (thermally initiated) RAFT in that it does not require a radical initiator (such as AIBN). Consequently, brushes can be grafted from initiator-functional surfaces 4 and bottle-brush architectures can be derived from macro-initiators without contamination by free chains. 18 The few reports of radically-made degradable brushes in the literature highlight the potential for biomedical applications. 19, 20 Riachi and coworkers 20 reported the surface-initiated atom transfer radical copolymerization of poly(ethylene glycol) methyl ether methacrylate (PEGMA) with the cyclic ketene acetal 5,6-benzo-2-methylene-1,3-dioxepane (BMDO). High BMDO feed ratios (10,25, and 50 mol-%) were used to compensate for the very low reactivity of BMDO in methacrylic copolymerizations. 2 The brushes were cleaved over the course of 30 days in the presence of acid (pH 3) or, more slowly still, base (pH 9). These unselective conditions also led to the cleavage of methacrylate side group esters and/or ester or siloxane groups within the surface-anchored initiators. More recently, Raj and coworkers 19 presented a pH-degradable molecular bottlebrush with potential for intracellular drug delivery applications. The multistep synthesis involved a core polymer carrying ATRP initiators connected via pH cleavable silyloxy linkers. The advantage was that degradation under intracellular conditions occurred at a predefined position directly between the brushes and the core to give fragments of predetermined size.
The combination of ATRP with TARO has not been reported but promises to widen the synthetic scope of both methods and to provide access to materials with tailored and selective degradability including such with brush-type architectures. Herein, we fill this gap by investigating the Cu(I)catalyzed ATRP of the thionolactone dibenzo[c,e]oxepine-5(7H)-thione (DOT) 8, 11 with acrylic comonomers. After observing and minimizing a side reaction-the Cu(I)-catalyzed dethionation of the thionolactone monomer which led to loss of DOT and some of the ATRP catalyst-we present the preparation of water-soluble bottlebrush copolymers able to shed their arms rapidly through selective persulfate oxidation.
Nuclear magnetic resonance spectroscopy (NMR) was done on a 400 MHz Bruker instrument in 5 mm NMR tubes. The residual non-deuterated solvent signals of CDCl3 (δH = 7.26 ppm) was used as reference.
Size exclusion chromatography (SEC) analysis was performed on a Viscotek GPC Max VE 2001 GPC. The system operated at 35 °C with three linear 7.5 mm × 300 mm Phenogel mixed-D columns connected to a refractive index detector. Tetrahydrofuran (THF) was used as mobile phase at a flow rate of 1 mL min −1 . The calibration of the system was based on the relative molar mass determination of a series narrow molecular weight distribution poly(methyl methacrylate) (pMMA) standards ranging from 0.88 to 1677 kg mol −1 and reported values are PMMA equivalent.
Dynamic Light Scattering (DLS) was performed on a Malvern Zetasizer Nanoseries instrument in glass cuvettes on polymer solutions with a concentration of 1 mg/mL. General procedure for normal ATRP. Copper (I) bromide (1 equiv) was suspended in solvent (DMSO, acetonitrile, toluene, 1 mL) followed by an addition of ligand Me6TREN (1 equiv). The mixture was degassed by purging with nitrogen for 25 minutes in a reaction tube with a sealed septum. The flask was opened with continued bubbling of nitrogen while vinyl comonomer and 6 DOT (in varying feed ratios as described in the main text) were added. The flask was quickly closed with the septum and degassed for a further 20 minutes. The mixture was then opened again with continued bubbling of nitrogen while methyl 2-bromopropionate (1 equiv) was added. The vial was resealed, degassed for a further 15 minutes, and placed into a preheated oil bath at 70 °C for 16 h. The polymerization was stopped by cooling and exposure to air. A sample (100 μL) was withdrawn, diluted with CDCl3 (500 μL) and analyzed by 1 H NMR spectroscopy to determine comonomer conversions. The crude polymerization mixture was filtered through basic alumina and dialyzed against MeOH (and, for water-soluble polymers) water, followed by drying.
General procedure for supplemental activation reducing agent (SARA) ATRP. Cu 0 wire (5 cm long, 1 mm diameter) was wrapped around a magnetic stir bar and treated with a solution of conc. aq HCl-MeOH (1:2 by volume) for 10 minutes, followed by washing with methanol and drying. PEGA (100 equiv), methyl-2-bromopropionate (1 equiv), CuBr2 (0.05 equiv), and the copper-clad stir bar were added to toluene (1 mL) in a septum-sealed vial. The mixture was purged with nitrogen for 20 min. The flask was opened with continued bubbling of nitrogen and Me6TREN (0.1 equiv) was added. Following additional degassing for 15 min, the mixture was heated, analyzed, and worked up as described above.
ATRP under anhydrous conditions. The vinyl comonomers and Me6TREN were dried over molecular sieves (4 Å). Toluene was either dried over molecular sieves (4 Å) or by distillation from sodium and storage over a potassium mirror. All glassware was dried in an oven overnight.
Separately, a mixture of toluene (0.5 mL), DOT (varying amounts), and methyl 2-bromopropionate (1 equiv) was similarly degassed in a septum-sealed vial. The DOT/initiator solution was then 7 transferred into the first flask through a canula followed by additional purging with nitrogen or argon for 20 minutes. The mixture was heated, analyzed, and worked up as described above.
General procedure for degradation of linear copolymers. Following previous studies, 10 copolymers were degraded by dissolving in a 1:1 (vol:vol) mixture of THF and 7 M ammonia in methanol (final polymer concentration 1 mg/mL) and stirring overnight at RT, followed by evaporation to dryness and analysis by SEC. The polymerization was quenched by exposure to air and cooling it to RT. Following analysis of a withdrawn sample by 1 H NMR spectroscopy, the reaction mixture was diluted with DCM and precipitated into diethyl ether-hexane (1:1 by volume).

Synthesis of bottlebrush copolymer (4).
The bottlebrush copolymer was prepared by classic ATRP following the above description by using the multi-initiator (1 equiv of bromides) and the polymer was isolated by dialysis as for linear PEGA copolymers.

Degradation of bottlebrush copolymer in aqueous solution.
To achieve fast degradation in aqueous solution, oxone (final concentration 10 mM) was added into a DLS sample (1 mL) containing bottlebrush copolymer (1 mg).

Results and Discussion
The synthesis of DOT-containing copolymers via ATRP is summarized in Scheme 1. Cu(I)Br (1 equiv) was used as catalyst for normal ATRP and methyl 2-bromopropionate (1 equiv) as initiator. To minimize catalyst inactivation (and loss of DOT) through undesired sulfur-copper interactions, tris[2-(dimethylamino)ethyl]amine (Me6TREN, 1 equiv) was chosen as a strongly complexing ligand known for its high ATRP activity. 22 The mixture of solvent, Cu(I)Br and Me6TREN was degassed by purging with nitrogen before DOT was added, followed by additional degassing, addition of the initiator, and a third degassing step, before the mixture was heated to 70 °C overnight. In an initial set of experiments, poly(ethylene glycol) methyl ether acrylate (PEGA, monomer Mn = 480 g/mol) was used as vinyl comonomer and DMSO as solvent. The combination of this polar solvent with Me6TREN as ligand causes some Cu(I) disproportionation. 23 While the 9 possibility of fast polymerization rates and low dispersities associated with this formulation have widely been appreciated, the mechanism has been the subject of debate. 24 This was in contrast to free radical and RAFT polymerizations, in which retardation is observed at higher DOT feed but copolymerizations with 30-40 mol-% DOT feed still proceed to reasonably high conversions. 6 Notably, when DOT was added into the polymerization mixtures, we observed the formation of brown and black precipitates, presumed to be copper sulfides leading to the irreversible loss of some of the ATRP catalyst and the cause for the absence of polymerization at higher DOT:Cu(I)Br feed ratios. For this reason, lower catalyst concentrations (equivalent to higher DOT:catalyst ratios) were not attempted despite literature reports on successful ATRP of similar systems with 100-times lower catalyst loads. 28 Herein, surprisingly, an unexpected side reaction was observed: the crude polymerization mixtures contained dibenzo[c,e]oxepane-5(7H)one (2, scheme 1), i.e., the (oxo)lactone, the synthetic precursor for DOT, see Figure 1. A control experiment containing DOT (5 equiv), CuBr (1 equiv), and Me6TREN (1 equiv) in DMSO (Table   1, entry 7, degassed as described above) gave 55% conversion to the lactone, while a similar   atmosphere and polar solvents were necessary for the reaction to proceed to completion and identified molecular oxygen and DMSO as oxygen sources. We therefore tried other solvents.
Unfortunately, a polymerization with a PEGA-DOT feed of 95:5 equiv in acetonitrile (Table 1, entry 9) similarly gave 44% lactone with 56% DOT incorporated and a low SEC-measured dispersity of Ð = 1.14. Similarly low dispersities (Ð = 1.10-1.20) were observed when toluene 13 was used as solvent (Table 1,   To further understand the copolymerization behavior, kinetics were measured for a PEGA-DOT feed of 95:5 using toluene under normal ATRP conditions (Table 1, entry 15). The conversion of the PEGA comonomer increased with time, reaching a value of 70% after 4.5 h, see Figure 3A.
Similar to the situation observed with RAFT and free radical polymerization, 6 Given that all above ATRPs proceeded in the presence of lactone, we investigated whether its presence led to chain transfer. A recent study concluded that chain transfer to the lactone was not significant during free radical polymerization. 30 In agreement, we found that the addition of up to 20 mol-% of lactone (2) to ATRPs of PEGA in toluene did not cause observable chain transfer, see Figure S1.
Despite the ability to make degradable copolymers with narrow size distributions, the seemingly unpreventable loss of typically 50% of the DOT feed and the inability to form polymers from formulations containing more than 15 mol-% of DOT meant that it was not possible to make copolymers containing more than ~7 mol-% DOT through the above methods. In a typical copolymerization, approximately 25 μmol of DOT were converted to lactone. Assuming a 1:1 thiocarbonyl-oxygen stoichiometry, the absence of dissolved oxygen, and using ideal gas law, this reaction required approx. 2.8 mL of air. It seemed unlikely that this volume of air remained in polymerization tubes of 10-15 mL volume following degassing by purging with nitrogen-a procedure used routinely in our group for radical polymerizations. Instead, we hypothesized that the oxygen source for the dethionations was adventitious water. Indeed, mercury-catalyzed dethionations have been reported to use water as the oxygen source. 31 Consequently, we performed ATRP under anhydrous conditions; the vinyl comonomers and Me6TREN were dried over molecular sieves and toluene was dried similarly or by distilling from sodium. To avoid exposure to ambient moisture, separately dried and degassed reagents were combined through canula transfer before the mixture was degassed further and then heated to 70 °C. Gratifyingly, a 95:5 PEGA-DOT copolymerization under anhydrous conditions led to 75% PEGA conversion, 10% residual DOT, 10% lactone formation, and 80% DOT conversion into the desired thioester backbone units, see Table 1

Conclusion
The Cu(I)-catalyzed atom transfer radical copolymerization of the thionolactone DOT is possible but is hampered by the dethionation of DOT in the presence of traces of oxygen or water.
This side reaction is fast and leads to the depletion of potentially large amounts of the thionolactone monomer. While Cu(I) has been used in the literature as a catalyst for similar dethionations, herein, the formation of black precipitates, believed to be copper sulfides, were observed and suggested irreversible loss of some ATRP catalyst. The lactone side product did not appear to interfere with the polymerization and, despite the above setbacks, degradable copolymers with low dispersities were formed, albeit with low (< 7 mol-%) thioester content without using strictly anhydrous conditions. Under anhydrous conditions, the formation of lactone could be minimized (to ≥ 5%) but not fully prevented. For the synthesis of well-defined linear DOT copolymers, we recommend using RAFT polymerization, for which side reactions (apart from retardation) have not been 22 reported. An advantage of ATRP lies in "grafting-from" polymerizations, demonstrated herein in the production of a water-soluble bottlebrush species that degraded rapidly through oxidative hydrolysis into smaller, water-soluble species.

Supporting Information.
Chain transfer experiments.

Corresponding Author
* corresponding author email address: p.roth@surrey.ac.uk