Vinyl copolymers with faster hydrolytic degradation than aliphatic polyesters and tunable upper critical solution temperatures

Vinyl polymers are the focus of intensive research due to their ease of synthesis and the possibility of making well-defined, functional materials. However, their non-degradability leads to environmental problems and limits their use in biomedical applications, allowing aliphatic polyesters to still be considered as the gold standards. Radical ring-opening polymerization of cyclic ketene acetals is considered the most promising approach to impart degradability to vinyl polymers. However, these materials still exhibit poor hydrolytic degradation and thus cannot yet compete with traditional polyesters. Here we show that a simple copolymerization system based on acrylamide and cyclic ketene acetals leads to well-defined and cytocompatible copolymers with faster hydrolytic degradation than that of polylactide and poly(lactide-co-glycolide). Moreover, by changing the nature of the cyclic ketene acetal, the copolymers can be either water-soluble or can exhibit tunable upper critical solution temperatures relevant for mild hyperthermia-triggered drug release. Amphiphilic diblock copolymers deriving from this system can also be formulated into degradable, thermosensitive nanoparticles by an all-water nanoprecipitation process.

V inyl polymers are attractive materials owing to their ease of synthesis and their broad diversity in terms of architectures, compositions, and functionalities, especially since the advent of reversible deactivation radical polymerization (RDRP) [1][2][3] . However, they are not degradable because of their carbon backbone, which creates environmental issues and severely limits their use for biomedical applications. Therefore, aliphatic polyesters are still the gold standards, especially for biomedical applications given their biocompatibility and degradability. Yet, the possibility to easily functionalize them and tune their structure and composition to obtain advanced materials are rather restricted. Consequently, combining the advantages of both polymer families to produce next-generation degradable materials is still an unmet need.
Yet, despite promising proofs of concept, the poor hydrolytic degradation of CKA-containing copolymers under physiological conditions remains a significant limitation. Indeed, even though their hydrolytic degradation under accelerated conditions is rapid, their degradation under physiological conditions usually takes from several months to a year to achieve at least a 50% decrease in molar mass, even with oligo(ethylene glycol) methyl ether methacrylate (OEGMA) as the main vinyl monomer [19][20][21] , which may be problematic for some (bio-related) applications. Therefore, such materials still cannot compete with the most popular polyesters like poly(lactic-co-glycolic acid) (PLGA) or even poly(lactic acid) (PLA). Their enzymatic degradation is also very limited as only copolymers comprising polycaprolactone (PCL)-like MDO units showed significant degradation in the presence of specific enzymes (e.g., lipases) 21 , whereas the bulky aromatic ring and/or the too high hydrophobicity of BMDO and MPDL drastically obstruct enzyme access 20,22 . In addition, for the design of materials with advanced physico-chemical, self-assembly, or stimuli-responsive properties 8 , the CKA is often used at least as a third comonomer and for a single goal; that is conferring degradability. This makes the synthesis more complex and, due to the high hydrophobicity of CKAs and/or side reactions during polymerizations, often has a detrimental impact on the targeted properties (e.g., the solubility of water-soluble copolymers 11 , colloidal stability, and particle size distribution of nanoparticles 23 , stimuli-responsiveness of materials 24,25 ). Therefore, allowing the CKA to combine two different properties would be important progress.
All these limitations are of utmost importance and represent serious obstacles to the development of advanced vinyl polymers for biomedical applications, such as thermoresponsive vinyl polymers which hold great promise for potential applications in bioengineering and nanomedicine 26 . They can exhibit either a lower critical solution temperature or an upper critical solution temperature (UCST); the latter being considered a very attractive feature for mild hyperthermia-triggered drug release. However, not only degradable, thermoresponsive vinyl polymers have received very little attention 17,22,[27][28][29][30][31][32] , especially UCST polymers 33,34 , but they all exhibit poor hydrolytic degradation and the synthesis of well-defined, UCST vinyl copolymers that can degrade in water has never been reported.
Herein, we report on a copolymerization system based on acrylamide (AAm) that circumvents the previously mentioned limitations associated with the rROP of CKAs. It enabled the synthesis of well-defined and cytocompatible vinyl copolymers exhibiting: (i) a rapid hydrolytic degradation in water or PBS, faster than that of PLA and even PLGA, which is unprecedented in the field of vinyl materials, and (ii) complete water-solubility or a tunable and sharp UCST in practically relevant conditions for which the CKA is an integral part of the thermosensitivity mechanism, thus greatly simplifying the system (Fig. 1). To demonstrate the great interest in these building blocks for biomedical applications, we also synthesized amphiphilic PEG-based diblock copolymers that were formulated into nanoparticles (NPs) by an all-water nanoprecipitation process owing to their UCST. It, therefore, avoided the use of organic solvents that are often problematic for pharmaceutical development, which is also unprecedented in the pharmaceutical field. These nanoparticles exhibited both UCST and LCST transitions, thus leading to doubly thermosensitive, degradable nanoparticles.

Results
Design rationale. The design of the copolymerization system relied on a simple structural analogy with poly(acrylamide-costyrene) (P(AAm-co-S)) copolymers which are known to exhibit a UCST in aqueous solution in the 50-62°C range, owing to reversible hydrogen bonds between the copolymer chains below the UCST and with water molecules above 35 . These copolymers however, received little attention compared to their acrylonitrilecontaining UCST counterparts; namely poly(acrylamide-coacrylonitrile) (P(AAm-co-AN)) copolymers 26,36 . Nonetheless, we postulated that substituting styrene with aromatic ring-containing CKA units, such as MPDL or BMDO, could be a straightforward route toward degradable, UCST copolymers in which the CKA would both confer degradability and govern the thermoresponsiveness, ideally in a much broader temperature range. It was also postulated that this copolymerization system would be compatible with RDRP to give access to well-defined architectures and that these would be more prone to hydrolytic degradation than Fig. 1 Design of hydrolytically degradable, thermosensitive vinyl copolymers. Synthesis of well-defined poly(acrylamide-co-cyclic ketene acetal) (P(AAm-co-CKA)) copolymers by reversible addition−fragmentation chain transfer (RAFT) copolymerization between AAm and CKAs, that exhibit tunable and biologically relevant upper critical solution temperature (UCST), and faster hydrolytic degradation than polylactide (PLA) and poly(lactic-co-glycolic acid) (PLGA).
any other CKA-containing vinyl copolymers synthesized so far given the very high water-solubility of AAm moieties that would promote efficient solvation of ester groups (Fig. 2).
Proof of concept with P(AAm-co-MPDL) and P(AAm-co-MDO) copolymers. To determine the most suitable aromatic ring-containing CKA to confer degradability and govern the thermosensitivity, the first copolymerizations were made with MPDL because of its open radical structure very close to that of styrene. P(AAm-co-MPDL) copolymers (P0-P4, Table 1) were obtained by RAFT-mediated copolymerization of AAm and variable initial molar fractions of MPDL (f MPDL,0 = 0-0.8) at 8 M in anhydrous dimethyl-sulfoxide (DMSO) for 16 h using 4-cyano-4-[(dodecylsulfanylthiocarbonyl)sulfanyl]pentanoic acid (CDSPA) as chain transfer agent (CTA) and azobisisobutyronitrile (AIBN) as the initiator. High monomer conversions were obtained and the copolymers exhibited molecular weights in the 8.3-28.3 kg mol −1 range with mostly low dispersities (Đ = 1.2-1.3) (Fig. 3a, c). By 1 H nuclear magnetic resonance (NMR), it was shown that the molar fraction of MPDL in the copolymer (F MPDL ) varied from 0.038 to 0.108 with 46% open MPDL units on average (Fig. 3b). Also, the more MPDL in the comonomer feed, the lower the molecular weight, as previously seen with other CKA/vinyl monomer pairs 23,37,38 .
The solubility of the different copolymers in water (10 mg mL −1 ) was then tested by transmittance measurements performed between 5 and 50°C (Fig. 3d). While PAAm (P0) and P(AAmco-MPDL) with F MPDL = 0.038 (P1) were fully water-soluble independently of the temperature, the copolymers with higher amounts of MPDL (P2-P4) showed a sharp UCST-type transition upon cooling. In addition, varying F MPDL from 0.043 to 0.108 allowed for fine-tuning the thermosensitivity, as shown by the increase of the cloud point (T cp ) from 15 to 38°C ( Supplementary Fig. 1). P(AAmco-MPDL) copolymers, P2, and P3 gave a UCST upon cooling but not upon heating because of the formation of stable aggregates. However, a reversible transition was observed for the copolymer with the highest MPDL content (P4), which gave the same cloud point near body temperature (T cp = 38°C) each time. Degradation of P(AAm-co-MPDL) copolymers was then performed under accelerated conditions (i.e., 5 wt.% aqueous potassium hydroxide (KOH) to probe the presence of open ester groups. Whereas P(AAm-co-MPDL) copolymer P1 showed a nearly constant M n after degradation because of its too low amount of open MPDL units, P(AAm-co-MPDL) copolymers P2-P4 led to a decrease in M n up to −31% as shown by size exclusion chromatography (SEC) (Fig. 3e and Supplementary Fig. 2). Although significant, this degradation however appeared lower than expected according to the theoretical M n after degradation values (M n, deg. theo ., Table 1). This can be explained not only by the limited amount of MPDL inserted in the copolymer due to the propensity of CKAs to hardly copolymerize with most acrylic monomers, but also by the significant fraction of ring-retained MPDL units.
These results confirmed our hypothesis as they showed that a simple RAFT-mediated copolymerization between AAm and an aromatic ring-containing CKA such as MPDL enabled the straightforward synthesis of well-defined, degradable, UCST vinyl copolymers owing to the presence of hydrogen bond donors (NH group of primary amide in AAm units) and acceptors (center of the phenyl rings in MPDL units) 39 .
Synthesis and evaluation of P(AAm-co-BMDO) copolymers. AAm was then copolymerized with BMDO as a second aromatic ring-containing CKA under similar conditions to: (i) assess the feasibility of well-defined, degradable, and UCST P(AAm-co-BMDO) copolymers and (ii) to determine the best copolymerization system before further physico-chemical and bio-related  (Table 1, P0-P4) in anhydrous DMSO initiated by AIBN at 70°C. Conversion = AAm conversion as determined by 1 H NMR; b 1 H NMR spectrum (300 MHz, DMSO-d 6 ) in the 0-9 ppm region of P(AAm-co-MPDL) copolymers P0-P4 (Table 1); c evolution of the SEC chromatograms of P(AAm-co-MPDL) copolymers P0-P4 as function of F MPDL ; d variation of the solution transmittance vs. temperature of P(AAm-co-MPDL) copolymers P2-P4 solution in water at 10 mg mL −1 (1°C min −1 , solid and dotted lines for cooling and heating, respectively); e evolution of the SEC chromatograms at a different time during the hydrolytic degradation under accelerated conditions (5 wt.% KOH) of P(AAm-co-MPDL) copolymer P4.
investigations. Beyond potential differences in terms of CKA insertion and amount of ring-opened units between MPDL and BMDO, we also suspected that the difference in position of the aromatic ring between those two monomers may impact the establishment of hydrogen bonds and therefore the thermosensitivity.
P(AAm-co-BMDO) copolymers (P9-P17, Table 2) were obtained at 0.8 M in anhydrous DMSO as copolymerizations performed between 2 and 8 M gave too highly viscous solutions in <1 h and greater proportions of closed BMDO (Supplementary Table 1 and Supplementary Figs. 8 and 9). By varying f BMDO,0 from 0 to 0.55, monomer conversions after 16 h were in the 96-67% range, and the more BMDO in the feed, the lower the conversion (Fig. 4a). The copolymers' molecular weights ranged from 18.6 to 6.1 kg mol −1 with rather low dispersities (Đ = 1.2-1.5) (Fig. 4c). Complete copolymerization kinetics was also performed for P13 via sample withdrawing at regular intervals to make sure of the linear growth of the copolymer with conversion ( Supplementary Fig. 10). Remarkably, it was shown by 1 H NMR that not only greater contents in BMDO could be introduced in comparison with MPDL at similar feed ratios (for instance, f BMDO,0 = 0.50 and f MPDL,0 = 0.80 led to relatively close F CKA of 0.102 and 0.108, respectively), but the average percentage of ring-opened BMDO units was much greater than with MPDL (89 vs. 46%, respectively), which suggests a greater susceptibility to hydrolysis (Fig. 4b). Given this strong benefit, we, therefore, decided to select the AAm/BMDO copolymerization system for further evaluation. Reactivity ratios of AAm/BMDO were determined to be r AAm = 13.02 and r BMDO = 0.23, using the nonlinear least-squares method (Supplementary Fig. 11) 40,41 . These values are rather similar to those reported for the copolymerization of BMDO with N-isopropylacrylamide (NIPAAm) (r NIPAAm = 7.31 and r BMDO = 0.11 at 120°C) 42 .
As expected, P(AAm-co-BMDO) copolymers also exhibited sharp UCST transitions upon both cooling and heating providing they contained enough BMDO (Fig. 4d). Indeed, PAAm (P9) and P(AAm-co-BMDO) with F BMDO = 0.017-0.068 (P10-P12) were fully soluble in water over the 0-100°C temperature range, whereas copolymers P13-P17 gave an increase of T cp from 25 to 52°C with F BMDO (Supplementary Fig. 12). Importantly, this enabled a range of UCST suitable for biomedical applications to be obtained since both room and body temperatures were covered simply by adjusting the BMDO content in the copolymer. Moreover, using BMDO in place of MPDL enabled fully reversible UCST with minimal hysteresis between cooling and heating to be obtained for all thermosensitive copolymers ( Fig. 4d). It also permitted a rather linear increase of T cp with F BMDO and a much broader range of T cp to be covered for a narrower range of CKA content (ΔT cp = 27°C for ΔF BMDO = 0.035 vs. ΔT cp = 23°C for ΔF MPDL = 0.065) (Fig. 4f). The latter point shows that fine-tuning the T cp is not at the expense of strong variations in the copolymer composition and its expected degradation. For comparison, non-degradable P(AAm-co-S) vinyl copolymers exhibited UCST behavior over a much narrower range of temperature and copolymer composition (ΔT cp = 12°C for ΔF S = 0.02) 35 .
Dynamic light scattering (DLS) was used to monitor the intensity average diameter (D z ) changes of the thermosensitive copolymers in water upon cooling and heating, and accounted for T cp values in excellent agreement with those obtained by transmittance measurements (Fig. 4e and Supplementary Fig. 13). Above the UCST, the copolymers were fully soluble and characterized by an average D z of~10 nm whereas, upon cooling, the average D z abruptly increased to~2-5 µm, demonstrating the formation of aggregates. Formation of such aggregates was also monitored by optical microscopy ( Supplementary Fig. 14).
To get further insight into the UCST behavior of this copolymerization system, the dependence of the cloud point on the molecular weight for a fixed f BMDO,0 value and on the copolymer concentration was also investigated. Additional copolymerizations with f BMDO,0 = 0.4 were first performed by targeting average degrees of polymerization, DP n, th , of 400 and 600 (P18 and P19, respectively, Supplementary Table 2) to be compared with P13 (DP n, th = 200). The higher DP n, th , the higher the molecular weight (M n, exp = 7.6-10.2 kg mol −1 , Đ = 1.4-1.8), but the lower the conversion and the BMDO content ( Fig. 5a and Supplementary Fig. 15). These results are in line with the unfavorable reactivity ratios that give gradient-type copolymers resulting in a gradual enrichment in CKA along the copolymer chain. Therefore, when the conversion is lower, so is the BMDO content. This has direct consequences on the UCST of the copolymers as the T cp was shifted toward lower temperatures (from~23 to~12°C as measured by transmittance and DLS), as the result of longer AAm sequences that promote polymer-water interactions ( Fig. 5b and c).
The dependence of the UCST on the copolymer concentration was then studied by measuring the T cp of aqueous solutions of P(AAm-co-BMDO) P13 at different concentrations ranging from 1 to 10 mg mL −1 (Fig. 5d). Similar to non-degradable AAm-based UCST copolymers 35,43,44 , the T cp gradually decreased with the copolymer dilution (from 25°C at 10 mg mL −1 to 20°C at 2.5 mg mL −1 ), until the copolymer loses its UCST at 1 mg mL −1 . At low copolymer concentration, the copolymer-copolymer intraand intermolecular hydrophobic interactions were less favored compared to copolymer-water hydrophilic interactions resulting in lower T cp values. Overall, targeting different DP n at a fixed f BMDO,0 value, and varying the copolymer concentration also independently offered the possibility to adjust the thermal response of P(AAm-co-BMDO) copolymers.
Hydrolytic and enzymatic degradation of the copolymers. The degradation of P(AAm-co-BMDO) copolymers was then evaluated under three different conditions:  Table 3).
Owing to their higher proportions in ring-opened BMDO than their MPDL-containing counterparts at similar compositions, degradation of P(AAm-co-BMDO) copolymers under accelerated conditions was much more pronounced, as shown by the significant shifts of the SEC traces toward lower M n even for copolymers with low BMDO contents (Table 2 and Supplementary Figs. 16 and 17). Indeed, 1.7 mol.% BMDO already led to 45% decrease in M n (P10), and very high degradations up to 90% decrease in M n were observed for higher F BMDO values, with a rather good agreement with the predicted values. 1 H NMR showed the nearly complete disappearance of the peak characteristics of proton h at 4.9-5.2 ppm in the alpha position to the ester group of BMDO, thus confirming the cleavage of the ester groups ( Supplementary Fig. 18).
Remarkably, P(AAm-co-BMDO) copolymers all exhibited extremely fast hydrolytic degradation in physiological conditions, whether they were fully water-soluble (P10-P12), or thermosensitive with a UCST below (P13) or above 37°C (P17). The decrease in M n , which was only governed by the BMDO content, was in the 21-58% range after only 24 h and in the 43-67% range after 7 days (Fig. 6a, b). Homogeneity of the degradation products was also demonstrated by the homogeneous shifts of the SEC chromatograms and the constant dispersity of the degradation products with time, except for P8 which exhibits the highest amount of MDO (F MDO = 0.44) (Supplementary Fig. 19). Not only these degradation kinetics were much faster than those of any CKA-containing vinyl copolymers reported so far (see Supplementary Fig. 20 for comparison with the literature), but they were also faster than those of traditional aliphatic polyesters like PLA and even PLGA under the same conditions ( Fig. 6b and Supplementary Fig. 21), which is unprecedented for vinyl copolymers. Considering that other CKA-containing hydrophilic copolymers (e.g., P(OEGMA-co-MPDL), P(GMA-co-BMDO)) 19,20 exhibit much slower hydrolytic degradation under physiological conditions than the AAm-counterparts, this highlights the crucial role of AAm in the very fast hydrolytic degradation of P(AAm-co-CKA) copolymers. This is likely explained by very high hydrophilicity of AAm moieties leading to optimal water uptake and efficient solvation of ester groups. These results represent an important achievement as, for the first time, a copolymerization system allowed to circumvent the very slow hydrolytic degradation in physiological conditions of CKA-containing vinyl copolymers and to surpass the hydrolytic degradation of traditional aliphatic polyesters. It is also interesting to note that the degradation was also successfully carried out in deionized water (pH~5.5) and resulted in a decrease in M n of almost 40% for P17 after only 3 days ( Supplementary Fig. 22). This fast degradation kinetics was also observed by UV measurements during five consecutive cooling and heating cycles of P(AAm-co-BMDO) copolymer P13 solution in water performed over a period of 7 h (Supplementary Fig. 23). The cloud point measured was notably affected from the first cycle to the fifth, as evidenced by a decrease of~1-2°C after each cycle, likely because of the on-going degradation of the copolymer. Interestingly, such tuning of the UCST transition during degradation is an interesting additional feature that had been shown on vinyl copolymers containing disulfide 30 or thioesters 34 linkages, but never before on CKA-based polymers.
Interestingly, despite their lack of thermosensitivity, P(AAmco-MDO) copolymers also exhibited rapid hydrolytic degradation in physiological conditions, whatever they were water-soluble (P6) or insoluble in water (P8) (Supplementary Figs. 24 and 25). Their degradation kinetics were similar to that of P(AAm-co-BMDO) copolymers with decrease in M n in the 65-75% range after 7 days. These results, therefore, broadened the application scope of the AAm/CKA copolymerization system as depending on the nature of the CKA and of its content, well-defined watersoluble, insoluble, or UCST copolymers with rapid hydrolytic degradation in physiological conditions/water can be readily synthesized.
PAAm, P(AAm-co-BMDO) copolymers P13 and P17, and P(AAm-co-MDO) copolymers P6 and P8 were also subjected to enzymatic degradation in PBS at 37°C in the presence of lipases from Candida antartica, a subclass of esterases. While the M n of PAAm stayed constant over time, degradation kinetics of P13 and  P17 were rather similar to those performed under physiological conditions and reached~60% after 7 days (Fig. 6c and Supplementary Fig. 26). We nonetheless suspected that, given the rapid hydrolytic degradation in PBS of these copolymers, it was difficult to determine the contribution of the enzymes and that the decrease in M n in Fig. 6c mainly reflected the contribution of hydrolytic degradation. As for MDO-containing copolymers, P8 was more rapidly degraded enzymatically than hydrolytically (especially within the first 3 days), likely owing to its high MDO content, whereas the opposite trend was obtained for P6 ( Supplementary Figs. 27 and 28).
In vitro cytotoxicity. Recognizing that it is essential to evaluate the safety of developed materials prior to any biopharmaceutical use, a small library of water-soluble or UCST, BMDO-containing copolymers with variable BMDO content (P9-P14 and P17, F BMDO = 0-0.128, Table 2) and M n (P13 and P18-P19, M n, exp = 7.6-10.2 kg mol −1 , Supplementary Table 2) was tested on three representative healthy cell lines to investigate any cytotoxic effects on the basis of cell viability assays and cell morphology observations ( Fig. 7 and Supplementary Figs. 29-32). The cell lines tested were: (i) murine fibroblasts (NIH/3T3), which are one of the most commonly used fibroblast cell lines; (ii) human umbilical vein endothelial cells (HUVEC), which are highly sensitive and give rapid response to external stimuli and (iii) murine macrophages (J774.A1), which are typical monocyte cells that play an important role in phagocytosis 45,46 .
Overall, all copolymers tested showed high cell viabilities (75-100%) by MTT(3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide) assay on the three cell lines at 0.01 and 0.1 mg mL −1 (Fig. 7a and Supplementary Fig. 29). Also, no clear trend was observed as neither the presence nor absence of a UCST transition, nor the BMDO content, nor the range of DP n, th tested seemed to influence the cell viability. Furthermore, varying the transition temperature over a wide range of values (T cp = 12-52°C) also did not affect the cytotoxicity. It is also worth mentioning that the 72 h-incubation time with cells (to allow at least two cell doubling times to perform relevant MTT assays) combined with the rapid hydrolytic degradation of the copolymers, suggests that neither the starting copolymers nor their degradation products were toxic to the cells. This point is key as it represents an important indication concerning the ultimate biocompatibility of the copolymers.
Cellular morphology observations after a 72 h-incubation with the different copolymers confirmed the cell viability results as no evidence of toxic effects were noticed compared to untreated cells or cells treated with unmodified PAAm (Fig. 7b and Supplementary Figs. [30][31][32]. Indeed, no difference in size, shape, cell density, or cell proliferation was observed whatever the copolymer, its concentration, or the cell line tested.
This cytotoxicity study, therefore, suggests that PAAm chemically modified by introducing BMDO units into its backbone did not adversely affect the cell viability and the morphology of three representative healthy cell lines.
Synthesis of block copolymers and application to doubly thermosensitive PEGylated nanoparticles. To illustrate the versatility of this copolymerization system, P(AAm-co-BMDO) copolymer P13 was first successfully chain extended with AAm via RAFT polymerization at 70°C for 2 h to produce a P(AAm-co-BMDO)-b-PAAm diblock copolymer (M n, exp = 18.2 kg mol −1 , Đ = 1.6). Despite slight broadening of the dispersity, gradual shift of the SEC traces toward higher molecular weights was shown. Its sacrificial P(AAm-co-BMDO) block was then partially degraded under hydrolytic conditions to yield a lower M n copolymer (M n, exp = 16.1 kg mol −1 , Đ = 1.5) as shown by SEC (Supplementary Fig. 33).
Macromolecular engineering was then further extended to the synthesis of P(AAm-co-BMDO)-b-POEGMA amphiphilic diblock copolymers for application in the formulation of degradable, PEGylated, UCST nanoparticles for drug delivery purposes. To determine the most efficient synthetic pathway, such copolymers were achieved by RAFT in anhydrous DMSO at 70°C by either: (i) chain extension of a POEGMA macro RAFT agent (M n, exp = 7400 g mol −1 , Đ = 1.1) by a 50:50 mixture of AAm and BMDO (P20) or (ii) chain extension of a P(AAm-co-BMDO) macro-RAFT agent (M n, exp = 4800 g mol −1 , Đ = 1.5, F BMDO = 0.069) by OEGMA (P21) (Supplementary Table 4 and Fig. 8a) Table 4), both copolymers successfully maintained their UCST properties with a little shift of the T cp values toward lower values close to ambient temperature (T cp~1 7-20°C) in comparison with a similar P(AAm-co-BMDO) copolymer (P14). This is likely due to the influence of the water-soluble POEGMA block, as already seen with other systems 44 . Interestingly, the presence of the POEGMA block also conferred an LCST transition to the copolymer at~73°C (Fig. 8b), and the three different solubility states can be reversibly reached upon temperature oscillation (Supplementary Movie 1).
By taking advantage of the UCST feature of the diblock copolymers, their formulation into well-defined, degradable nanoparticles was achieved through an organic-solvent-free, all-water nanoprecipitation process in the presence of 0.1 wt.% Pluronic as a surfactant. It consisted in adding an aqueous solution of P(AAm-co-BMDO)-b-POEGMA copolymer prepared at 25°C (T > UCST) into an aqueous solution at 5°C (T < UCST) under stirring (Supplementary Fig. 36). This nanoprecipitation procedure can therefore operate entirely in water, which is very advantageous compared to the classical nanoprecipitation technique that requires the use of a watermiscible organic solvent (e.g., acetone, tetrahydrofuran (THF), etc.), whose removal may often be incomplete. DLS measurements revealed that well-defined nanoparticles were obtained in both cases with average diameters of~200 nm and narrow particle size distributions (Supplementary Table 4). The nanoparticles' formation and colloidal characteristics were shown to be reversible upon oscillating the temperature between 5°C (insoluble state, T < UCST, nanoparticles with P(AAm-co-BMDO) core), 40°C (soluble state, UCST < T < LCST) and 85°C (insoluble state, T > LCST, nanoparticles with POEGMA core) up to at least three consecutive cycles (Fig. 8c, d and Supplementary Figs. 37 and 38), thus leading to degradable, doubly thermosensitive 47,48 copolymer nanoparticles. These variations in number-, intensity-and volume-average diameters ( Supplementary Figs. 37 and 38), confirmed the different transitions observed by UV measurements. Transmission electron microscopy (TEM) analysis confirmed the formation of spherical and narrowly dispersed nanoparticles (Fig. 8e and Supplementary  Figs. 39 and 40), exhibiting diameters in the 60-80 nm range, which correspond to the diameters in the dry state.
The evolution of their colloidal properties was monitored by DLS upon storage at 37°C and 4°C (Supplementary Fig. 41). A sharp decrease in average diameter from 180 nm down to 50 nm was obtained after 3 days at body temperature, thus attesting for their disintegration (likely via concomitant disassembly and hydrolytic degradation). On the other hand, at 4°C, their average diameter remained almost constant during storage, due to a slowing down of the degradation process (but not its stopping as shown by the increase of the particle size distribution).

ARTICLE
To expand the field of possibilities in terms of applications, the UCST transition was also adjusted close to body temperature simply by tuning the POEGMA chain length and the composition of the P(AAm-co-BMDO) block. This was achieved by targeting a shorter POEGMA macro-CTA (M n, exp = 3200 g mol −1 , Đ = 1.33), in order to promote higher UCST values, followed by chain extension with AAm/BMDO (f BMDO,0 = 0.55) to yield a final M n, exp = 7700 g mol −1 ; Đ = 1.5, with F BMDO = 0.126 (Supplementary Table 5). This copolymer (P22) exhibited a UCST transition in the 40-42°C range as shown by transmittance and DLS measurements (Supplementary Fig. 42 and 43), which could be relevant for mild hyperthermia therapy. Interestingly, a progressive shift of its T cp to lower values during degradation under physiological conditions was observed until 16 days when it reached 3°C (Supplementary Fig. 44).

Discussion
In this study, we discovered that the RAFT-mediated rROP of AAm with different CKAs (MPDL, BMDO and MDO) represented a straightforward route toward the synthesis of a broad range of mostly well-defined P(AAm-co-CKA) copolymers that exhibited fast and tunable degradation in physiological conditions (PBS, pH 7.4, 37°C), that even surpassed that of PLA and even PLGA in the same conditions. With aromatic ring-containing-CKAs, the resultant copolymers even showed tunable and sharp UCST transitions in the~10-50°C range, which could be of high interest for drug delivery applications. The UCST transition could be readily and independently adjusted by modifying the CKA content in the copolymer, the copolymer concentration, and the targeted average degree of polymerization. Since the copolymerization was successfully applied to both MDO and MPDL/ BMDO, the degradation can therefore be decoupled from the thermosensitivity, which is also advantageous depending on the targeted application. Moreover, preliminary cell viability assays showed that they had a good cytocompatibility on three different representative healthy cell lines, representing a key result considering biomedical applications.
To demonstrate the robustness and the broad applicability of this synthetic approach, POEGMA-b-P(AAm-co-CKA) diblock copolymers were synthesized, whose UCST allowed their formulation by an innovative all-water nanoprecipitation process, that avoided the use of organic solvent, into narrowly dispersed nanoparticles of~200 nm in diameter suitable for drug delivery applications. Owing to the LCST of the POEGMA block, these nanoparticles even exhibited both UCST and LCST transitions, and can be considered as the first example of doubly thermosensitive, degradable nanoparticles.
Considering all its important features and advantages, we believe that this two-in-one copolymerization system, in which the CKA confers both efficient degradability and thermosensitivity, could lead to very useful macromolecular building blocks that would greatly expand the current and somewhat limited arsenal of degradable materials. It can therefore offer exciting perspectives for a wide range of different applications ranging from degradation of hydrophilic plastic-based materials such as PAAm which is a key polymer in this field 49 , to hyperthermia-mediated drug delivery and tissue engineering.
Analytical methods NMR spectroscopy. NMR spectroscopy was performed in 5 mm diameter tubes in DMSO-d 6 at 25°C. 1 H-NMR spectroscopy was performed on a Bruker Avance 300 spectrometer at 300 MHz. The chemical shift scale was calibrated based on the internal solvent signals (δ = 2.50 ppm for DMSO-d 6 ). Data were processed with MestReNova 11.0.4 software.
SEC. SEC was performed at 60°C using two columns in series from Agilent Technologies (PL PolarGel-M, 300 × 7.5 mm; bead diameter 8 µm; molar mass range 1000-5,00,000 g mol −1 ) preceded by a guard column from Agilent Technologies (PL PolarGel-M, 7.5 × 50 mm; bead diameter 8 µm) and a triple detection system (Viscotek TDA/GPCmax from Malvern) with a differential refractive index detector, low and right-angle light scattering detectors and a differential viscometer detector. The eluent was DMSO with 100 mM LiBr and 0.36 wt.% of 2,6-di-tertbutyl-4-methylphenol (BHT) as a marker at a flow rate of 0.7 mL min −1 . The system was calibrated using poly(methyl methacrylate) (PMMA) standards (peak molar masses, M p = 540-342,900 g mol −1 ) from Agilent Technologies. This allowed the determination of the number-average molar mass (M n ), the weightaverage molar mass (M w ) and the dispersity (Ð = M w /M n ). All samples were filtered over 0.22 µm PTFE filters prior to injection. Data were collected and processed with OmniSEC 4.0 software.
UV-Vis spectroscopy. Light transmittance (%) of samples was measured using a Lambda 25 UV/VIS spectrometer equipped with a PTP 1 + 1 Peltier system for temperature control (PerkinElmer), at a wavelength of 500 nm, a cell path length of 10 mm and under magnetic stirring. Samples were prepared at 10 mg mL −1 in deionized water and placed in a quartz cuvette. The measurements were carried out by first cooling the solution from the used highest temperature (T >> UCST) at a constant rate of 1°C min −1 followed by reheating the solution back to the starting temperature at the same rate. The inflection point of the transmittance curve was considered as the UCST cloud point. It was graphically determined by the maximum of the first derivative of the cooling/heating curves. Data were collected with Winlab 6.0.3.0730 software and processed with Data Processor and Viewer (DPV) 1.00.100.0010 software.
DLS. Nanoparticles' intensity average diameters (D z ) and particle size distributions were measured by DLS with a Nano ZS from Malvern equipped with a 4 mW He −Ne laser (633 nm wavelength) at a fixed scattering angle of 173°. Similar to UV-Vis measurements, samples were prepared at 10 mg mL −1 and placed in a quartz cuvette. The cooling/heating measurements were taken at an interval of 1°C starting from T >> UCST and the solution was equilibrated at each temperature for 60 s prior to the measurements. The inflection point of the intensity average diameters curve was considered the cloud point (UCST). It was graphically determined by the maximum of the first derivative of the cooling/heating curves. Data were processed with Malvern-Zetasizer 7.12 software.
Optical microscopy. Optical images of copolymer solution in water were captured with a Leitz Diaplan microscope equipped with a Coolsnap ES camera (Roper Scientific). The samples for optical microscopy were prepared by placing a drop of the copolymer solution (10 mg mL −1 ) at T < UCST and T > UCST on a glass microscope slide prior to examination. Data were collected and processed with RS Image V1.9.2 software.
TEM. Grids were glowing and discharged before use. 5 μL of nanoparticle suspensions (1.67 mg mL −1 ) kept at 4°C were deposited for 5 min on copper grids covered with formvar-carbon film. The excess solution was blotted off using filter paper. Samples were then stained for 30 s using phosphotungstic acid (2%, w/v) also stored at 4°C. Then the excess solution was removed using filter paper. The grids were then analyzed using a JEOL JEM-1400 operating at 80 kV. Images were acquired using an Orius camera (Gatan Inc, USA). Data were collected and processed with Zen 2.6 (blue edition) software. Sizes were measured by ImageJ 1.53a.
Synthesis of poly[oligo(ethylene glycol) methyl ether methacrylate] (POEGMA) macro-CTA. A typical synthesis of POEGMA 23 macro-CTA was conducted as follows: in a pre-dried 50 mL round bottom flask, fitted with a rubber septum and a magnetic stir bar, a mixture of OEGMA (3.68 g, 0.012 mol), CDSPA (1 eq., 0.24 mmol, 0.097 g) and AIBN (0.25 eq., 0.059 mmol, 9.6 mg) was dissolved in anhydrous acetonitrile (25 mL). The solution was bubbled with dry argon to remove dissolved oxygen for 15 min at room temperature and then immersed in a preheated oil bath at 70°C for 5 h. The solution was then rapidly cooled under air. Acetonitrile was removed and the resulting polymer solution was precipitated once in excess of cold mixture of 1:1 of diethyl ether and petroleum spirit. The POEGMA 23 macro-CTA (M n, exp = 7400 g mol −1 , Đ = 1.1) obtained was then dried under a high vacuum until constant weight.
Degradation procedures Accelerated degradation. In a 20 mL vial equipped with a magnetic stirrer, 50 mg of the desired copolymer was dissolved/dispersed in 2.5 mL of deionized water at 30°C. After sonication and complete solubilization/dispersion, 2.5 mL of potassium hydroxide solution (5 wt.%) in deionized water was added. The mixture was then stirred for 1 h at room temperature at which time the solution turned completely transparent. The solution was then quenched by adding an aqueous solution of HCl (1 M). The resulting solution was then freeze-dried overnight to yield a white powder. The degraded product was then analyzed by SEC chromatography.
Hydrolytic degradation. In a 7 mL vial, 20 mg of the desired copolymer was dissolved/dispersed in 2 mL of PBS (pH 7.4) or deionized water and the solution was mechanically stirred in an orbital shaker oven (IKA KS4000i control) set at 150 rpm and thermostated at 37°C. At specific time intervals (i.e., 1, 3, and 7 days), samples of 0.5 mL were withdrawn and freeze-dried. The degradation products were then analyzed by SEC.