Dual Thermal- and Oxidation-Responsive Polymers Synthesized by a Sequential ROP-to-RAFT Procedure Inherently Temper Neuroinflammation

This study is about multiple responsiveness in biomedical materials. This typically implies “orthogonality” (i.e., one response does not affect the other) or synergy (i.e., one increases efficacy or selectivity of the other), but an antagonist effect between responses may also occur. Here, we describe a family of very well-defined amphiphilic and micelle-forming block copolymers, which show both oxidative and temperature responses. They are produced via successive anionic ring-opening polymerization of episulfides and RAFT polymerization of dialkylacrylamides and differ only in the ratio between inert (N,N-dimethylacrylamide, DMA) and temperature-sensitive (N,N-diethylacrylamide, DEA) units. By scavenging Reactive Oxygen Species (ROS), these polymers are anti-inflammatory; through temperature responsiveness, they can macroscopically aggregate, which may allow them to form depots upon injection. The localization of the anti-inflammatory action is an example of synergy. An extensive evaluation of toxicity and anti-inflammatory effects on in vitro models, including BV2 microglia, C8D30 astrocytes and primary neurons, shows a link between capacity of aggregation and detrimental effects on viability which, albeit mild, can hinder the anti-inflammatory potential (antagonist action). Although limited in breadth (e.g., only in vitro models and only DEA as a temperature-responsive unit), this study suggests that single-responsive controls should be used to allow for a precise assessment of the (synergic or antagonist) potential of double-responsive systems.

Figure S1. A. 1 H and B. 13 C NMR spectra of DOT (red spectra, black assignments) and DOTAc (black spectra, red assignments).C. IR spectrum of DOTAc.Red arrows point to the absorptions related to the stretching vibrations of methyl (as at 2865 cm -1 ) and carbonyl groups (1686 cm -1 ).

1.2SI Synthesis of propylene sulfide (PS)
460 g (6 mol, 1.25 equiv per propylene oxide (PO)) of ammonium thiocyanate was dissolved in 800 mL of deionized water obtaining a concentration of 0.58 g/mL (7.6 M) and the resulting solution was degassed with argon gas for 1 hour.It was thereafter introduced in an Atlas HD reactor (Syrris Ltd., Royston, United Kingdom), comprising a jacketed reactor, jacket, and S4 reactor temperature probe, 50W heating probe, overhead stirrer, and an inlet for dry argon gas.
Under an argon atmosphere, 400 mL of degassed DCM was added while stirring, to form an emulsion with a 2:1 water:DCM (v/v) ratio.280 g (4.8 mol) of PO was subsequently added continuously via a dropping funnel at a rate that allowed the reactor setup to control the temperature between 28 and 32 °C (2.3 mL/min in our setup).The final PO concentration with respect to the aqueous phase was 6 M (0.35 g/mL).The reaction was left to run for 16 hours after initial PO addition.The organic phase was thereafter distilled out of the reactor under reduced pressure (300 mbar) and at 35 °C, collecting it in a dry ice-cooled flask.The distillate was dried over Na 2 SO 4 followed by fractional distillation to separate DCM and PS between 80 °C (DCM-rich phase, confirmed by ATR-IR) and 120 °C (PS-rich phase) under an argon atmosphere.The average yield was 58% by weight, with an average purity of 99 mol% (only impurity being DCM).ΔH= 47±3.4 kJ/mol.

1.3SI Optimization of end-capping conditions
Six potential RAFT-active end-cappers were employed: benzyl bromide (BB, a primary halide), ethyl 2-bromopropionate (BP), ethyl α-bromophenylacetate (BPA), 2bromopropionamide (BPAM), 2-bromopropionitrile (BPN)) and ethyl α-bromoisobutyrate (BIB).The data are summarized in Table S1 and the procedures leading to quantitative endcapping and monomodal molecular weight distributions (use for further experiments) are highlighted in red.The general experimental procedures are always those described in the main text, except for the parameters mentioned in Table S1.Notes: A) BPAM.In run 23, the 1 H NMR spectrum indicated quantitative end-capping, the C=S signal was missing in the 13 C NMR spectrum and GPC chromatogram showed multimodal distribution.By doubling all reagents (TBP, CS 2 and BPAM) as in run 25, GPC secondary S6 peaks were reduced, and by further replacing Na with DBU (as the thiolate counterion in the end-capping) as in run 26, the C=S signal appeared in the 13 C NMR spectrum.However, in no experiment monomodal MW distribution were obtained and this end-capper was therefore abandoned.
B) BPN.The end-capping was not successful when using Na as a counterion (runs 19, 21,   22), but also when DBU was used instead, a multimodal MW distribution was observed (run 20).Therefore, BPN was also excluded from further evaluation.
C) BIB.End-capping tried in THF at room temperature was completely unsuccessful with multimodal MW distribution (data not shown).The reaction was therefore attempted in DMF, adding about 10% degassed water, and using longer reaction times, all factors thought to promote the SN 1 reaction of the sterically hindered end-capping agent.However, quantitative end-capping was not achieved in any of the conditions assessed (runs 26-29).
D) BB, BP, and BPA.For these monomers, conditions were found to obtain quantitative endcapping and monomodal MW distribution.For BB (benzylic halide), this meant a 1: 1.5 : 4 : 5 thiol/TBP/CS 2 /BB molar ratio (run 2).For BPA (benzylic and electron-poor, but also very sterically hindered), the same conditions yielded a multimodal distribution (run 12), and a monomodal one could only be obtained using double the concentration of all reagents (run 15).
For BP (secondary bromide), under most conditions the end-capping was relatively S7 satisfactory, but it was only using higher excesses of reagents (1: 3 : 8 : 10 thiol/TBP/CS 2 /BP molar ratio), double concentration of all and DBU as a counterion that a monomodal MW distribution was obtained.In some experiments, all reagents were doubled in concentration; these experiments are here highlighted as those with higher PS concentration, but please note that the PS / thiol stoichiometric ratio remains constant (=20) throughout all the table.
b Na was always the counterion during polymerization.In some cases, 2.2 equivalents of acetic acid were added at the end of the polymerization, followed by 2.3 equivalents of DBU, i.e. first thiolates and any residual alcoholate were protonated, then only thiols were deprotonated again with DBU leading to a large, organic counterion and a 'naked' thiolate.
c The initiator peak at 3.70 ppm in 1 H NMR was used as a reference to calculate the end-capping yield (mol%) from CH 2 of BB at 4.65 ppm; CH 3 of BP at 1.30 ppm, CH 3 of BPA at 1.30 ppm, CH of BPN at 3.50 ppm, CH of BPAM at 3.50 ppm, CH 3 groups of BIB overlapping with CH 3 of PPS at 1.30 ppm.
d The presence of multiple peaks is typically attributed to low end-capping yield, which allows thiolates to eventually produce disulfides, thereby multimerizing the macromolecular structure and generating multimodal molecular weight distributions.
e No resonance in the 1 H NMR spectrum can be associated to the end-capper structure.
f Not recorded, because it was already known that the end-capping performed poorly.
g DBU was always used in 1.15:1 molar ratio with thiols.
h In these experiments, 2.5 eq.s of azidobenzene (in relation to thiols) were added at the end of the PS polymerization and before the addition of CS 2 , to avoid potential side reactions via quenching TBP.
1.4SI Characterization of the selected macroRAFT agents S9 UV-spectroscopy was employed to quantify the number of C=S groups per chain, as well as to ensure the complete end-capping of the polymer chains.For that, a commercially available model molecule (bis(carboxymethyl)trithiocarbonate) was employed, measuring its extinction coefficient at 310 nm (ε=11,487 L/mol) and using the latter to calculate the molar concentation S10 of trithiocarbonates in THF solution.The molar concentration of the macromolecules in solution was obtained by dividing the polymer concentration in g/L by the values obtained

Mn
through GPC or MALDI-ToF.
According to the spectra presented in Figure S4, BP-PMA features the predicted two C=S groups per chain, but BB-PMA and BPA-PMA would appear to have around three.The result for the latter two, however, is difficult to rationalize on the basis of the synthetic approach: the three PMAs differ in the electrophile added in the last step, but the reaction leading to the trithiocarbonate (PPS tholates + CS 2 ) is identical; therefore, they should all have the same number of C=S groups.It is therefore likely that the extinction coefficient of aromaticcontaining trithiocarbonates is simply different (higher) than that of a fully aliphatic one, i.e. bis(carboxymethyl)trithiocarbonate is an unsuitable model for PMA and BPA-PMA .

1.5SI Optimization of MALDI-ToF analytical procedures
Our group previously reported MALDI-ToF analysis of PPS polymer end-capped with ethyl-2-bromoacetate using 2-(4-hydroxyphenylazo)benzoic acid (HABA) as matrix in the presence of silver trifluoroacetate (AgTFA) as a cationizing agent.Under the same conditions, however, no polymer signal acquisition was obtained with the polymers synthesized herein except for the BP-NEG, where the signal intensity was very low compared to the signals acquired at lowmolecular weight region (Figure S5).We ascribed this to the different functionalities present at the polymer chain ends which might influence the ionization of the polymers.A more S12 hydrophobic matrix dithranol gave clearer polymer chain populations in the spectra where the best spectrum was obtained in the absence of a cationizing agent (Figure S6).These results confirmed that the polymer chain end group influences the ionization of a polymer.To identify the effect of trithiocarbonate at the polymer chain end, we performed a series of MALDI-ToF analysis of BP-PMA using different matrices, namely 1,4-dicyanobenzene (DCB), 2,5dihydroxybenzoic acid (DHB), HABA, α-cyano-4-hydroxycinnamic acid (CHCA), or dithranol at varying polymer-to-matrix ratios in the presence or absence of NaTFA or AgTFA.
Polymer chain population was observed only when HABA (only in the absence of a cationizing agent) or dithranol (except for the presence of AgTFA) was used.An increase in the polymer signal was observed when the polymer-to-matrix ratio was increased up to 1:19 and when linear ionization mode was employed but the fragment signals still overlapped with the polymer (Figure S7).Dithranol provided the best signal acquisition of BP-PMA with a negligible number of fragments in the spectrum when used in the absence of a cationizing agent and with relatively lower amount of matrix unlike HABA (Figures S8 and S9).The same was observed for BP-NEG (Figure S6) where the only difference between the chain end functionalities is the presence of trithiocarbonate group in BP-PMA.This indicates that the presence of trithiocarbonate does not significantly affect ionization of polymers.

Figure S2 .
Figure S2. A. In situ calorimetric curve during the reaction.B. 1 H NMR spectrum of PS.

Figure
Figure S4.UV-spectra of the PPS-macroRAFT agents (PMAs).The absorbance values at 310 nm were used to calculate the concentration of C-S groups, according to Beer-Lambert's law.Please note that the peak location is the same despite the structural differences in the trithiocarbonate (a benzyl residue on BB-PMA (red), an aromatic β ester on BPA-PMA (light gray), and an aliphatic β ester on BP-PMA (black)), but the extinction coefficient of the band may and probably is affected by them.

Figure S10 .Figure S11 .Figure S12 .Figure S13 .Figure S16 .
Figure S10.MALDI-ToF analysis of BB-PMA with HABA as a matrix in the presence or absence of cationizing agents at different polymer-to-matrix ratios (linear positive ion mode).

Figure S17 .S27Figure S18 .
Figure S17.Dependency of the 500 nm optical density of PSE on temperature, at different concentrations in deionized water

Figure S20 .
Figure S20.Left.The 48h exposure of neurons to all polymers measurably decreased the viability in comparison to 24 h, although it only approached 50% of the MTT signal (not normalized against protein content) at the highest concentrations (  0.5 mg/mL) of the polymers with the largest DEA content (PSE and PSM1E3).Right.Livedead stains (calcein / propidium iodide) con cortical neurons exposed for 24h to all polymers at five different concentrations.

Table S1 .
Reaction conditions for the end-capping of PPS with RAFT-active groups