Immolative Polymers

SELF-IMMOLATIVE POLYMERS

1. Introduction

In recent years, there has been significant interest in the development of stimuli- responsive polymers for a wide range of applications. For example, thermo- responsive polymers can serve as valves in microfluidic devices (1), whereas polymers responding to intrinsic biological stimuli such as enzymes, or changes in pH or redox potential, can be used to selectively deliver drug molecules to diseased sites in vivo (2). There are several established mechanisms by which polymers can respond to stimuli. The stimulus can trigger a change in the charge state or solubility of the polymer backbone or pendant functional group along the polymer backbone (Fig. 1a). For example, polyamines undergo protonation–deprotonation in a pH-dependent manner, which can result in water-soluble materials at neu- tral and acidic pH, and water-insoluble materials at basic pH (3,4). Alternatively, a stimulus can result in the cleavage of pendant groups along the polymer backbone (Fig. 1b). For example, pendant cyclic acetal groups undergo hydrolysis at acid pH, resulting in the transformation of the pendant hydrophobic groups to hydrophilic ones, changing the polymer from water insoluble to water soluble (5). Stimuli can also trigger cleavage of the polymer backbone and ultimately its degradation into small molecules (Fig. 1c). For example, polymers with pH- sensitive acetals (6), reduction-sensitive disulfides (7), or photochemically cleav- able o-nitrobenzyl ester moieties (8) in their backbones have been reported. A limitation to most of the approaches described above is that multiple stimuli-mediated events are required to cause significant changes in the polymer properties. For example, multiple stimuli-mediated cleavages of pendant groups or backbone moieties are required to change the solubility of a polymer or to break it down to small molecules. While large changes in chemical environment and high concentrations of stimuli are easily achieved in the laboratory, in real applications the changes in environmental conditions are generally more subtle and the concentrations of stimuli are much lower. Therefore, there is significant interest in the development of approaches that can amplify the responses of materials to stimuli. Self-immolative dendrimers were developed in 2003 almost si- multaneously by the groups of de Groot (9), Shabat (10), and McGrath (11) with the aim of amplifying responses to stimuli. These molecules employed the self- immolative spacer concept initially developed for prodrugs in which cleavage of a trigger moiety initiates a spontaneous intramolecular reaction such as an elimination or cyclization, to release a drug molecule (Fig. 2) (12–15). The linkage of multiple self-immolative spacers sequentially in branched form led to dendrimers

Fig. 1. Schematic representations of the mechanisms by which polymers can respond to stimuli: (a) Change in solubility of the polymer backbone or pendant groups, (b)cleavage of pendant groups from the backbone, and (c) cleavage of the polymer backbone.

Fig. 2. (a) Schematic of a prodrug employing a self-immolative spacer and trigger; (b) general scheme for an elimination spacer; (c) general scheme for a cyclization spacer. that could fragment, releasing multiple molecules from the dendrimer periph- ery upon cleavage or activation of a trigger moiety at the dendrimer focal point (Fig. 3a) (9–11). Since their conception, a wide range of self-immolative dendrimers have been developed, with potential applications ranging from drug delivery vehi- cles (16,17) to chemical sensors (18–20). Self-immolative spacers have also been combined in oligomeric form through stepwise synthesis to enable the release of terminal as well as multiple pendant groups upon triggering (Fig. 3b) (21–25). However, limitations of these systems include their tedious, stepwise synthesis, as well as issues of steric hindrance in the case of dendrimers, which limits the number of branching layers (ie, generations) that can be prepared and thus the extent of signal ampliﬁcation. SELF-IMMOLATIVE POLYMERS 3

Fig. 3. Cleavage of a trigger moiety by a stimulus initiates degradation of a self- immolative: (a) dendrimer; (b) oligomer, including loss of pendant groups; and (c) polymer.

The preparation of linear self-immolative polymers (SIPs) through one-step polymerization reactions offers a possibility to overcome the limitations of mul- tistep dendrimer and oligomer syntheses, as well as the steric hindrance issues associated with dendrimers. Their design is similar to that of the dendrimers in that activation or cleavage of a trigger moiety at the polymer terminus typically initiates a cascade of reactions that results in depolymerization (Fig. 3c). Indeed the concept of depolymerization has been known for decades in the context of polymers with low ceiling temperatures (Tc). For example, polyformaldehyde was developed by DuPont as the first engineering plastic, but required stabiliza- tion via acetate end-capping (26). It has a ceiling temperature of 120°Cinits uncapped form, above which the entropy gained through depolymerization over- rides the relatively small enthalpic gain of polymerization, but this is increased to >200°C through end-capping (27). More recently, it was shown by Darensbourg and co-workers that polycarbonates prepared from the alternating copolymerization of epoxides and CO2 depolymerize to the corresponding polymerization monomers or cyclic carbonates, depending on their chemical structure and the reaction conditions (28–30). It was proposed that the regeneration of monomers through depolymerization would be particularly attractive for the recycling of plastic waste (31). Despite many decades of knowledge on polymer Tc and depolymerization, the use of stimuli-responsive end-caps in combination with backbones that depolymerize by sequences of well-defined reactions has enabled a new level of control over this degradation process, allowing SIPs to be exploited for signal amplification and stimuli-responsive materials. Several recent review arti- cles have described the development of this field (32–35). Described here is the 4 SELF-IMMOLATIVE POLYMERS

Fig. 4. Mechanism of the 1,6-elimination reaction. development of several classes of self-immolative linear polymers including polycarbamates, polycarbonates, polythiocarbamates, polythiocarbonates, polyacetals, and poly(benzyl ether)s, including their syntheses and depolymerization mechanisms. Their application in functional materials including sensors, capsules, nanoscale polymer assemblies, and microscale pumps is described. The depolymerization kinetics is also summarized, as this is a key feature that distin- guishes the degradation of SIPs from that of traditional biodegradable polymers. Finally, the current state-of-the-art for the ﬁeld and future outlook is discussed.

2. Poly(benzyl carbamate)s

2.1. Introduction of Poly(benzyl carbamate)s as SIPs and Their Ap- plication as Enzyme-Responsive Sensors and Probes. The ﬁrst linear SIP backbone was introduced in 2008 by Shabat and co-workers (36). It was based on the 4-aminobenzyl alcohol spacer that had previously been widely exploited in prodrugs (12–14) and previous self-immolative dendrimers (18) and oligomers (22,24). When incorporated into electron-withdrawing groups such as carbamates, the aniline nitrogen is insufﬁciently electron donating to undergo an elimination reaction, but upon deprotection, revealing the electron-donating aniline, the molecule undergoes a 1,6-elimination reaction to generate an azaquinone methide and release the substituent on the benzylic position (Fig. 4). The released azaquinone methide can further react with surrounding nucleophiles such as water, regenerating aromaticity. The approach of Shabat and co-workers involved the preparation of a phenyl carbamate derivative of 4-aminobenzyl alcohol (36). This derivative was quite stable at room temperature, but underwent polymerization in the presence of dibutyltin dilaurate (DBTL) in N,N-dimethylformamide (DMF) at high temperature (100°C), followed by a reaction with an alcohol- functionalized end-cap molecule to provide the target self-immolative polycarbamate (Fig. 5). Cleavage of the end-cap revealed the aniline, triggering depolymerization via alternating 1,6-elimination and loss of CO2. In this initial work of Shabat and co-workers, 4-hydroxy-2-butanone was used as the end-cap (36). In the presence of bovine serum albumin (BSA), this SELF-IMMOLATIVE POLYMERS 5

Fig. 5. Synthesis of a self-immolative poly(benzyl carbamate) and its depolymerization following end-cap cleavage. end-cap was removed via β-elimination, triggering depolymerization (Fig. 6a). To detect depolymerization and enable use of the polymer as a turn-on fluorescent sensor for BSA, the 4-hydroxybenzyl alcohol monomer was modified with an acrylate substituent ortho to the amine. When the amine was functionalized as a carbamate, this monomer exhibited only weak fluorescence, but upon release of the amine, the monomer exhibited strong fluorescence emission at 510 nm. Thus, the production of the free amine upon depolymerization in the presence of BSA led to a significant increase in fluorescence emission. In addition, while the pendant carboxylic acid group of the acrylate could be protected as a t-butyl ester during polymerization, cleavage of these t-butyl protecting groups from the resultant polymer provided many ionizable carboxylic acid groups, imparting water solubility. In subsequent work, Shabat and co-workers modified the design to incorporate 4-nitroaniline as a pendant group on each monomer (37) (Fig. 6b). Depoly- merization was activated by the cleavage of the 4-hydroxy-2-butanone end-cap by piperidine, and a subsequent 1,6-elimination of the depolymerized monomers released the 4-nitroaniline reporter molecules. In this case, the solubility of the polymer required that the depolymerization be performed in an organic solvent, where it is relatively slow. This was addressed through the synthesis and polymerization of a dimeric monomer with a protected carboxylic acid group on one unit and a 4-nitroaniline reporter molecule on the other. This polymer was end- capped with a phenylacetamide moiety, designed for cleavage by penicillin-G ami- dase (PGA). Following deprotection of the carboxylic acid moieties, the resulting polymer was water soluble and could be triggered by PGA to depolymerize, releasing the reporter molecules. Because the azaquinone methide intermediate generated upon depolymerization is highly reactive, in addition to water molecules, it can also react with other nucleophiles. Shabat and co-workers have used this reactivity for the activity-linked labeling of enzymes (Fig. 6c) (38). The phenylacetamide and 4-hydroxy-2-butanone end-caps described above were used to impart sensitivity 6 SELF-IMMOLATIVE POLYMERS

Fig. 6. (a) A water-soluble poly(benzyl carbamate) with pendant carboxylic acid groups undergoes depolymerization in the presence of BSA to release fluorescent monomer units; (b) depolymerization of a poly(benzyl carbamate) releases 4-nitroaniline reporter molecules; and (c) activity-linked labeling of enzymes through the reaction of azaquinone methide species with nucleophilic groups on proteins (adapted with permission from reference (38). Copyright 2009 American Chemical Society). SELF-IMMOLATIVE POLYMERS 7 to PGA and a catalytic antibody Ab38C2, respectively. Following small molecule model studies, it was demonstrated that the labeling of both PGA and Ab38C2 could be achieved after cleavage of the SIPs by these enzymes. After labeling, PGA did not exhibit a significant reduction in its activity. However, the activity of Ab38C2 did decrease as the concentration of the SIP probe in the reaction increased, suggesting that labeling of the active site lysine ε-amine interfered with catalytic activity. It was also demonstrated that SIPs could provide enhanced lev- els of labeling while preserving higher catalytic activity in comparison with self- immolative monomers or oligomers. This was attributed to the gradual break- down of the SIP over seconds to minutes, during which time the SIP could diffuse farther from the active site, allowing azaquinone methide species to be released in the vicinity of protein nucleophiles whose modification would not affect catalytic activity. Despite this possibility for polymer diffusion, when both an activating and nonactivating protein were present during the reaction, the labeling of the activating protein was eightfold higher than that of the nonactivating protein. 2.2. Application of Poly(benzyl carbamate)s in Paper-Based Sen- sors. In addition to the initial enzyme sensors described by Shabat, self- immolative poly(benzyl carbamate)s have also been used in a variety of other sensor devices. For example, Phillips and co-workers incorporated poly(benzyl carbamate) oligomers with H2O2-sensitive aryl boronate end-caps as phase-switching agents in quantitative time-based assays (39). The principle behind this design is that upon depolymerization the SIP changes from a hydrophobic, water- impermeable layer to hydrophilic, water-soluble degradation products, allowing water to wick through the layered, paper-based device, dissolving food coloring in a subsequent layer (Fig. 7). The resulting brightly colored solution provides a simple visual readout for the device, the only required measurable being time for the signal to be produced, which depends on the concentration of H2O2. This aspect makes these devices promising for applications in resource-limited environments including the developing world. It was found that the use of oligomers improved the detection limit of the device by 4 orders of magnitude to 6 nM H2O2, in comparison with an analogous device using a small molecule (40). This was attributed to the amplification effect afforded by the depolymerization mechanism. It was proposed that the sensitivity could be further improved by using longer polymers, but only if they depolymerized more rapidly than the residence time in the device. The same team has recently expanded this assay to employ a cascade of events involving aptamers and enzymatic reactions to ultimately trigger cleavage of the aryl boronate end-cap. This has enabled the detection of inorganic ions including Pb2+ and Hg2+, with the possibility to expand to other analytes including small molecules, enzymes, and other inorganic ions (41,42). 2.3. Enhancing the Rate of the 1,6-Elimination Reaction. While the sensors developed by Phillips and co-workers successfully demonstrated the potential utility of the self-immolative poly(benzyl carbamate)s, they also highlighted one of their limitations. The degradation rate of these SIPs is relatively slow in polar environments and is even slower or may not occur at all in environments with low dielectric constant. They suggested that as depolymerization is proposed to occur via less aromatic transition states resembling azaquinone methides, it should be possible to reduce the energy penalty and thus increase the depolymerization rate by two possible approaches (43). One approach involved 8 SELF-IMMOLATIVE POLYMERS

Fig. 7. (a) Schematic of the design of a phase-switching, time-based assay; (b) observed read-out on the actual device; and (c) chemical structure of the SIP employed in this device. (Adapted with permission from reference (39). Copyright 2013 American Chemical Society.) reducing the aromaticity of the parent structure (Fig. 8a) by replacing the ben- zene ring with a naphthalene (Fig. 8b). A second approach involved the addition of a methoxy group to the aromatic ring to increase electron density and raise the highest occupied molecular orbital (HOMO) (Fig. 8c). This approach was success- ful with the naphthalene derivative providing a 113-fold enhancement and the methoxy derivative providing a 143-fold enhancement in the rate of depolymerization. Thus, this work provided important guidelines on the design of rapidly degradable poly(benzyl carbamate)s. 2.4. Thermal Triggering of Poly(benzyl carbamate) Depolymeriza- tion. In addition to the end-caps described above, Boydston and co-workers have recently used a bicyclic 1,2-oxazine end-cap to afford thermal triggering SELF-IMMOLATIVE POLYMERS 9

Fig. 8. Chemical structures of 4-aminobenzyl alcohol derivatives with increased rates of 1,6-elimination: (a) parent structure, (b) naphthalene derivative, and (c) methoxy derivative.

Fig. 9. (a) Proposed mechanism of thermally-activated end-cap cleavage and (b)synthesis of a block copolymer containing a SIP and a heat-sensitive linkage between the blocks. of self-immolative polycarbamates (44). As shown in Figure 9a, it was anticipated that heating would result in cycloreversion of the oxazine to an unstable carbamoylnitroso intermediate, which following hydrolysis would rapidly decar- boxylate to generate the free amine, initiating depolymerization (45). Poly(N,N- dimethylacrylamide) (PDMA) was synthesized by atom transfer radical polymerization (ATRP) from a pentamethylcyclopentadiene-based initiator and then reacted with hydroxyurea end-capped SIP under oxidative conditions to afford a diblock copolymer via in situ oxidation to the nitroso followed by cycloaddition reaction (Fig. 9b). It was shown that the resulting block copolymer underwent depolymerization in a temperature-dependent manner in 9:1 deuterated dimethyl sulfoxide (DMSO-d6):D2O. In comparison, depolymerization was slower for a control polymer with a nonresponsive end-cap, though some depolymerization was observed for the control polymer at higher temperatures, suggesting that hydrolysis is also involved at greater than 60°C. Trapping studies with cyclohexadiene further supported the role of the carbamoylnitroso intermediate and thus the proposed thermolysis mechanism. Thus, in addition to chemical stimuli it is also possible to use heat to trigger depolymerization of SIPs. 10 SELF-IMMOLATIVE POLYMERS

Fig. 10. (a) Preparation of microcapsules from a self-immolative poly(benzyl carbamate) and (b) changes in capsule shell morphology under different conditions. (Adapted with permission from reference (46). Copyright 2010 American Chemical Society.)

2.5. Application of Poly(benzyl carbamate)s in Stimuli-Responsive Microcapsules. Through the incorporation of pendant t-butyldimethylsilyl (TBDMS) protected hydroxyl groups along the poly(benzyl carbamate) backbone, Moore and co-workers prepared cross-linkable SIPs (46) (Fig. 10a). t- Butyloxycarbonyl (Boc) and ﬂuorenylmethyloxycarbonyl (Fmoc) end-caps were used, providing sensitivity to acid and base, respectively. After removal of the TBDMS groups, the hydroxyl groups were converted to isocyanates by a reaction SELF-IMMOLATIVE POLYMERS 11

PBC-b-PDMA

PDMA-b-PBC-b-PDMA Triggered disruption via self-immolative O O O depolymerization O O O N N H H O n N

Trigger cleavage Visible light trigger (PB1 and PB2) O O O CO O O O 2 H2N.. N H O n N NO2 UV light trigger (PB3) Self-immolative depolymerization S PDMA S O OH Reductive n n HO +1 CO2 +1 H N millieu trigger (PB4) 2 O N

Fig. 11. Synthesis of diblock copolymers containing self-immolative blocks with light- and reduction-sensitive linkers and their self-assembly into vesicles. End-cap cleavage leads to the release of cargo from the vesicles. (Adapted with permission from reference (47). Copyright 2014 American Chemical Society.) with excess 2,4-toluene diisocyanate (2,4-TDI). An emulsion of water with gum arabic as a surfactant and viscosity modifier, and ethyl phenylacetate as the organic phase containing the isocyanate-functionalized polymers, was prepared. In- terfacial cross-linking was performed using butanediol. Microcapsules with sizes ranging from 5 to 40 μm containing ethyl phenylacetate in their cores were produced. Upon triggering with either HCl (Boc capsules) or piperidine (Fmoc capsules), the capsules released their core contents over a period of 24–48 h. In contrast, capsules prepared from a non-self-immolative control polymer released negligible core contents over this period. In addition, electron microscopy revealed that the triggered capsules became cracked and deflated, whereas control capsules were unaffected by the acidic or basic triggering conditions (Fig. 10b). These results suggest that such capsules are promising for potential applications such as drug delivery and self-healing materials. 2.6. Incorporation of Poly(benzyl carbamate)s into Amphiphilic Block Copolymers for Aqueous Self-Assembly. Poly(benzyl carbamate)s have also been incorporated into block copolymers for self-assembly. As the parent poly(benzyl carbamate) is water insoluble, its linkage to a hydrophilic block provides an amphiphilic block copolymer that can self-assemble in aqueous solution to form nanoassemblies. Liu and co-workers synthesized the poly(benzyl carbamate) by the previously reported method (36) and incorporated end-caps including perylen-3-yl methanol, 2-nitrobenzyl alcohol, and diethanol disulfide, which are visible light-, UV light-, and reduction-responsive end-caps, respectively (Fig. 11) or a noncleavable benzyl alcohol end-cap (47). The remaining benzyl alcohol termini were then functionalized with a reversible addition-fragmentation chain transfer (RAFT) agent. RAFT polymerization was then used to grow a 12 SELF-IMMOLATIVE POLYMERS hydrophilic PDMA block, resulting in block copolymers with hydrophilic fractions on the order of 60–70 wt%. Self-assembly of the polymers in aqueous solution was then investigated. The assemblies were studied by transmission electron microscopy (TEM), scanning electron microscopy (SEM), confocal laser scanning microscopy, and dynamic light scattering (DLS). It was found that these block copolymers formed vesicles with diameters ranging from 200 to 580 nm. This behavior was despite their relatively high hydrophilic fraction in comparison to the expected hydrophilic vol- ume fraction of 25–45 wt% for vesicle-forming block copolymers (48). This was attributed to the strong hydrogen-bonding interactions between the carbamate groups. Upon end-cap cleavage with light or reducing conditions, the vesicles were shown by microscopy, NMR spectroscopy, and size exclusion chromatogra- phy (SEC) to depolymerize and disintegrate. The release of 4-aminobenzyl alcohol was also demonstrated. Liu and co-workers also exploited the capabilities of vesicles to incorporate hydrophilic payloads in their aqueous cores and hydrophobic payloads in their membranes (47). Using the reduction-sensitive disulfide system, they encapsulated hydrophilic doxorubicin as its HCl salt and hydrophobic camptothecin. They demonstrated selective release of both drugs in the presence of the reducing agent glutathione. Encapsulation and release of the photosensitizer eosin was also investigated. Furthermore, using different combinations of the light- and reduction-responsive vesicles, OR, AND, and XOR logic gate-type programmed enzymatic reactions were constructed. 2.7. Encapsulation and Release of Poly(benzyl carbamate)s from Viral Capsids. In contrast to using the SIP in the shell of vesicles or capsules, Cornelissen and co-workers have encapsulated water-soluble self-immolative polycarbamates with pendant acrylic acid groups (as in Fig. 6a above) into the cores of Cowpea Chlorotic Mottle Virus (CCMV) capsids (49). The SIP was capped with a 5-methoxy-2-nitrobenzyl carbamate to afford depolymerization in response to UV light. Upon irradiation with 350 nm light, depolymerization occurred, resulting in a significant increase in the fluorescence emission at 490 nm indicative of depolymerization, and the resulting monomers were released through the capsid pores. DLS and TEM suggested that the capsid underwent a morphological change and shrinkage upon depolymerization, a phenomenon that was not observed for capsids containing nondegradable poly(styrene sulfonate) that were irradiated. The presence of Mg2+ ions was found to stabilize the capsules with re- spect to this morphological change, likely through binding to the pores, yet it did not prevent depolymerization or the release of monomers. Overall, this concept offers an alternative strategy for the loading and noninvasively triggered release of small molecules from capsids.

3. Other Polycarbamates, Polycarbonates, Polythiocarbamates, and Polythiocarbonates

3.1. Development of Polycarbamates Incorporating Cyclization Spacers. In addition to the self-immolative poly(benzyl carbamate)s that SELF-IMMOLATIVE POLYMERS 13

Fig. 12. (a) Synthesis of a polycarbamate based on 4-hydroxylbenzyl alcohol and N,N- dimethylethylenediamine as well as its depolymerization via a sequence of reactions involving cyclization, 1,6-elimination, and loss of CO2.(b) Chemical structure of an amphiphilic self-immolative block copolymer. (c) TEM image of assemblies of this block copolymer formed in aqueous solution (scale bar = 100 nm). (Adapted with permission from reference (51). Copyright 2009 American Chemical Society.)

depolymerize via 1,6-elimination reactions and loss of CO2, another important category of SIPs is those depolymerizing by cyclization reactions. As described below, these cyclization reactions can be used to tune the depolymerization rate when used in combination with the 1,6-elimination, CO2 elimination sequence, and they also reduce the generation of potentially toxic quinone methide species (50) that arise from the 1,6-elimination. Using an activated monomer based on 4-hydroxylbenzyl alcohol and N,N-dimethylethylenediamine, DeWit and Gillies prepared a self-immolative polycarbamate capped with a Boc group (51). As shown in Figure 12a, in the presence of base as well as 5 mol% of Boc-protected monomer as an end-cap, the amines on the monomers reacted with the 4-nitrophenyl carbonates on other monomers to provide a polycarbamate. Upon removal of the Boc group and incubation in pH 7.4 phosphate buffer:acetone (3:2), the SIP underwent depolymerization by a sequence of cyclization, 1,6-elimination, and loss of CO2 to afford N,N - dimethylimidazolidinone, 4-hydroxybenzyl alcohol, and CO2. In the same study, DeWit and Gillies also demonstrated that it was possible to incorporate a poly(ethylene glycol) (PEG) end-cap using a 4-nitrophenyl carbonate-activated PEG as an end-cap in the polymerization (Fig. 12b). This 14 SELF-IMMOLATIVE POLYMERS

Fig. 13. Chemical structure of (a)UVand(b) NIR light-responsive polycarbamates. resulted in an amphiphilic block copolymer that self-assembled in aqueous solution to afford micellar-type nanoaggregates (Fig. 12c) (51,52). Upon cleavage of a single ester group between the PEG and SIP blocks, depolymerization was initiated, ultimately resulting in disintegration of the nanoaggregates. Encapsu- lation of a model payload nile red was demonstrated, and its release throughout the depolymerization process was suggested by a significant decrease in its fluorescence, as the dye is well known to aggregate upon its release into the aqueous environment resulting in significant quenching of fluorescence. In 2012, Almutairi and co-workers introduced UV light- and near infrared (NIR) light-sensitive o-nitrobenzyl and 4-bromo-7-hydroxycoumarin end-caps, respectively, to this polycarbamate backbone (Fig. 13) (53). Complete depolymerization of the resulting polymers was demonstrated by SEC and 1H NMR spectroscopy following relatively short irradiation times. Using a single emulsion procedure, these SIPs were used to prepare nanoparticles. Following UV or NIR light irradiation. the nanoparticles were disrupted, resulting in a burst release of encapsulated nile red. Furthermore, the depolymerization products were found to exhibit minimal cytotoxicity to RAW 264.7 macrophage cells in a 3-(4,5-dimethyl- 2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay, suggesting the potential of these stimuli-responsive nanoparticles for drug delivery applications. 3.2. Increasing the Depolymerization Rate by Tuning the Cycliza- tion Spacer – SIP Backbones Incorporating Carbonates and Thiocar- bonates. Although the self-immolative polycarbamates both with and without cyclization spacers have been successfully synthesized and demonstrated to depolymerize in response to a variety of stimuli for different applications including sensors and materials for controlled release, their depolymerization rates are relatively slow. This was highlighted as a potential limitation in the flow-through sensors developed by Phillips (39). In addition, in the work of Gillies (51) and Al- mutairi (53), the polycarbamates containing the N,N-dimethylethylenediamine spacer required days to weeks to depolymerize, depending on the conditions. To address this limitation, Phillips and co-workers have tuned the chemical structures of 1,6-elimination spacers to afford rapid elimination and slow background hydrolysis as described above (43,54). DeWit and Gillies have also developed 4- aminobutyric acid spacers that cyclize in seconds in pH 7.4 aqueous buffer (55), but they have not yet been incorporated into polymer backbones due to synthetic challenges. However, Gillies and co-workers have incorporated two simple modifi- cations to the N,N-dimethylethylenediamine spacer, to afford rapidly depolymerizing polycarbamates based on 4-hydroxybenzyl alcohol (56). As shown in Figure 14a, the replacement of one amino group of N,N-dimethylethylenediamine with SELF-IMMOLATIVE POLYMERS 15

Fig. 14. (a)and(b): Chemical structures and depolymerization mechanisms for SIPs similar to those in Figure 12, but where the depolymerization rate is accelerated by (a) replacement of the carbamate with a carbonate and (b) replacement of the amine nucle- ophile in the cyclization reaction with a thiol; (c) chemical structure and depolymerization mechanism for a SIP that degrades entirely by cyclization reactions. an oxygen converted a backbone carbamate into a carbonate, which was more electrophilic and facilitated the cyclization reaction. Through this modification, the time required for depolymerization was reduced from days for the parent polycarbamate to hours for the poly(carbamate-carbonate). Replacement of the other amino group with a thiol provided even slightly faster depolymerization (Fig. 14b). 3.3. Development of an SIP Backbone Depolymerizing Entirely by Cyclization Reactions. Currently, there are mixed data concerning the potential toxicity of quinone methide and azaquinone methide depolymerization products (50). For biomedical applications in particular, the potential for these reactive species to react irreversibly with proteins is a significant concern (38). To address this potential issue, Gillies and co-workers developed a SIP derived from N,N-dimethylethylenediamine and 2-mercaptoethanol, which degraded entirely by cyclization reactions, without the generation of reactive quinone methides (Fig. 14c) (57). This polymer, as well as the one shown in Figure 14b, had reduction-sensitive disulfide end-caps, with the potential to be cleaved under physiological conditions such as within cells where the concentration of the biological reducing agent glutathione is 0.5–10 mM in comparison with 2–20 μMin the extracellular environment (58,59), or within hypoxic tumor tissue (60), where 16 SELF-IMMOLATIVE POLYMERS

Fig. 15. General synthesis and depolymerization of a polyacetal. it is proposed to be about fourfold higher than in normal tissue. The depolymerization of this backbone occurred relatively slowly over a period of about 10–14 days, suggesting that it would be most useful for applications requiring a slow release of payload.

4. Polyacetals

4.1. Polyacetals as Low Ceiling Temperature (Tc) Polymers. An- other important category of SIPs is the polyacetals, including polyphthalaldehyde (PPHA) and its copolymers, as well as polyglyoxylates. Polyacetals depolymerize due to their relatively Tc (26). For example, non-end-capped PPHA is well known to have a ceiling temperature of approximately –40°C and therefore depolymer- izes spontaneously at room temperature (61–63). This can be attributed to the unstable hemiacetal termini, which induce rapid head-to-tail depolymerization (Fig. 15). However, with proper end-capping, polyacetals can be stable well above the Tc of the uncapped polymer (61–63). Because of the high dipole moment of the carbonyl bond of aldehydes, they are susceptible to ionic polymerization by both anionic and cationic mechanisms, with the required polymerization temperature dependent on the Tc. 4.2. Introduction of Polyphthalaldehydes as SIPs. PPHA and its depolymerization chemistry have been known for decades. For example, it was used by Frechet,´ Ito, and Wilson in photoresist chemistry, where backbone cleavage of a stable, end-capped PPHA initiated by a photoacid generator resulted in complete depolymerization (63,64). However, it was much more recently that Phillips and co-workers recognized the potential of using stimuli-responsive end-caps with PPHA, to afford a new level of control over PPHA depolymerization and thus materials that were selective to various chemical signals. In their initial work, they used n-butyllithium (n-BuLi) to anionically polymerize o-phthalaldehyde (OPA) at –80°C over a period of 10 days and end-capped it with functional moieties responsive to stimuli including Pb(0), fluoride, or a control, nonresponsive end-cap (Fig. 16a) (65). After end-capping, the PPHAs were stable for at least 15 h in tetrahydrofuran (THF), but once the end-caps were removed by the desired conditions, the polymers completely depolymerized in minutes. They also prepared stimuli-responsive plastics by patterning a cylinder of the fluoride-responsive polymer in control polymer (Fig. 16b). Upon exposure to fluoride, the polymer depolymerized to produce a cylindrical hole in the plastic (Fig. 16c). One limitation to expanding the scope of application of PPHAs was the long polymerization time of more than 10 days noted above. Using a modification of a protocol developed by Hedrick and co-workers (66), Phillips and co-workers SELF-IMMOLATIVE POLYMERS 17

(a) O Ο O O Responds to: Ο Pd (0) n Ο O O OHC CHO Anionic polymerization End-capping Ο O O n Ο Control n

tBu O O O Si Me Me F n (b) Patterned plastic sheet (c)

tBu O O O Si Me Me Chemical trigger signal Chemical signal n that reacts with the TBS trigger

Plastic with altered features After Before

O O O Control

Fig. 16. (a) Synthesis of PPHA end-capped with moieties responsive to different stimuli. (b) Preparation of a stimuli-responsive plastic by the patterning of a cylinder of fluoride- responsive PPHA in a control non-responsive PPHA. (c) Exposure to fluoride results in the production of a cylinidrical hole in the plastic. (Adapted with permission from reference (65). Copyright 2010 American Chemical Society.) developed a scalable and reproducible synthesis of PPHA using 1-tert-butyl- 5 5 2,2,4,4,4-pentakis(dimethylamino)-2λ ,4λ -catenadi-(phosphazene) (P2-t-Bu) as a nitrogen-base catalyst (67). This procedure allowed PPHA to be synthesized in 3 h, instead of multiple days, and allowed various different end-caps to be incorporated at either the initiating or terminal end of the polymer. In addition, control of PPHA molecular weight could be achieved based on the amount of initiator alcohol added. The purity of the OPA monomer was found to be critical to obtain good yields of polymer. An additional development in this study was the demonstra- tion that depolymerization of PPHA could occur rapidly in the solid state upon exposure of a photochemically sensitive PPHA to UV light, even in the absence of a solvent. 4.3. Application of Polyphthalaldehydes in Microscale Pumps. Phillips, Sen, and co-workers have also explored the application of PPHAs in single-use self-powered microscale pumps that could be turned on by specific stimuli (68). A TBDMS end-capped PPHA insoluble film served as the basis of this technology. Exposure of the polymer film to fluoride ions as the signal resulted in end-cap cleavage and depolymerization to more than 100 monomers per 18 SELF-IMMOLATIVE POLYMERS

Fig. 17. (a) Design principle of a self-powered micrometer-scale pump based on PPHA. (b) Variation of the design incorporating a β-glucuronidase-sensitive small molecule that produces fluoride ions to trigger the depolymerization. polymer chain, thereby amplifying the signal and and creating a concentration gradient that pumped fluids and insoluble particles away from the bulk polymer by a diffusiophoretic mechanism (Fig. 17a). The pumping speed of this type of mi- cropump ranged from 0.1 to 11 μms−1, depending on the concentration of the sig- naling molecule. Furthermore, the pump was capable of moving particles around corners and over distances of approximately 5 mm. It was also demonstrated that the PPHA pump could be tuned to be responsive to different analytes, including enzymes (67,68). For example, a β-D-glucuronidase sensitive glucose derivative with a self-immolative spacer that released fluoride ion upon glucose cleavage by the enzyme was incorporated, such that released fluoride would turn on the pumping system (Fig. 17b). In general, when end-capped SIPs are used as solid-state materials, such as in the micrometer-scale pumps described above, it is necessary that the end- caps be accessible for cleavage in the liquid and therefore accessible at the solid– liquid interface. Phillips and co-workers used the micrometer-scale pump as a test system to evaluate the effect of end-cap polarity and polymer length on end- cap accessibility, as this system provided the functional output of tracer particle movement that could be measured (69). The β-D-glucuronidase system described above was used. Silyl ether derivatives with varying hydrophilicities including a t-butyl group, a hydrophilic oligo(ethylene glycol), or a 30-carbon-long hydro- carbon were used, and varying molecular weights of PPHA were prepared with each end-cap. It was found that for short-to-moderate length polymers (eg, Mn of 8–35 kg mol–1), faster pumping and thus faster PPHA depolymerization were observed for the more hydrophilic silyl end-caps. This was attributed to the increased concentration of silane end-cap at the polymer film surface for the more hydrophilic end-caps, which was supported by analysis of the films by x-ray photo- electron spectroscopy. However, for longer PPHA (eg, 60 kg mol–1), the pumping SELF-IMMOLATIVE POLYMERS 19

(a) 3–10 wt% polymer in CHCl3 10 wt% PVA

PVA 5 wt% + FITC-DEX

10 wt% PVA (b)

Fig. 18. (a) Schematic illustrating the preparation of PPHA microcapsules by a microfluidic flow-focusing technique. (b) SEM images illustrating the destruction of the capsule wall upon exposure to fluoride. (Adapted with permission from reference (70). Copyright 2013 American Chemical Society.) speeds were similar for all end-caps. At such lengths, the rate of pumping was thought to no longer be limited by the rate of end-cap cleavage, but rather the low end-cap content, and therefore the depolymerization time itself. Overall, this study provided a new understanding of how to tune the accessibility of end-caps at solid–liquid interfaces and ultimately the pumping speed. It is envisoned that these “turn-on” micro-pumps will have applications in microanalysis, microfluids, and diagnostic devices. 4.4. Application of Polyphthalaldehydes in Stimuli-Responsive Mi- crocapsules. Phillips, Weitz, and co-workers have also prepared microcapsules from TBDMS end-capped PPHA (70). As for the previously described polycarbamate microcapsules developed by Moore and co-workers (46), the aim was to utilize the amplification effect afforded by SIPs to increase the sensitivity of the capsules to stimuli. In this study, a flow-focusing fabrication technique was used, which is ideal for the preparation of capsules containing aqueous cores under mild conditions without the requirement of chemical reactions for the incorporation of the polymer into the shell wall. The polymer was dissolved in chloroform and microfluidic flow-focusing was used to encapsulate fluorescein isothiocyanate (FITC)-labeled dextran (Fig. 18a). Poly(vinyl alchol) (PVA) was added to both the core and external aqueous solutions to balance the osmotic pressure. The thickness of the shell wall was controlled by varying the flow rates of the different phases and the PPHA concentration. The resulting microcapsules had smooth surfaces and diameters of approximately 150 μm. It was found that exposure to fluoride resulted in the formation of holes in the capsule wall and the release of FITC-dextran (Fig. 18b). Capsules with thinner walls released 20 SELF-IMMOLATIVE POLYMERS

FITC-dextran more rapidly. Those composed of shorter PPHA chains also released their contents more rapidly for the same reasons described above for the micrometer-scale pumps. 4.5. Functionalizable Polyphthalaldehyde Copolymers. While PPHA is promising for many applications due to its rapid depolymerization, even in solid state, as well as the commercial availability of OPA as a polymerization monomer, the chemical modification of this polymer is challenging because of the sensitivity of the polymer backbone and because functionalizable derivatives of the monomer are neither commercially available or easy to access synthetically. To address this limitation and expand the potential ultility of PPHA in new applications, Kaitz and Moore have explored the copolymerization of OPA with substituted benzaldehydes to introduce functional groups for further modification of this polymer (71). They hypothesized that while the ceiling temperature of polybenzaldehyde itself is too low to enable polymerization under accessible conditions due to its low enthalpic gain relative to entropic cost of polymerization, the more exother- mic nature of hydrate formation with electron-deficient benzaldehyde derivatives would translate into higher polymerizability. Through a series of copolymerization experiments, they found that benzyladehyde derivatives with Hammett values higher than 0.92 were incorporated into the polymers (Fig. 19a). The incorporation of functional groups such as halides or aldehydes provides sites for the subsequent functionalization of the polymers. For example, the aromatic halides underwent Stille and Sonogashira couplings to provide alkene or alkyne groups for cross-linking or “click” functionalization (Fig. 19b). Pendant aldehyde groups could be reduced to alcohols and reacted with isocyanates to afford cross-linking or used as initiation sites for the synthesis of polylactide (Fig. 19c). While this work was performed with acetate or trichloroacetyl carbamate end-caps, making the polymers stimuli-responsive mainly through backbone cleavage, it should be feasible to readily extend this approach to PPHA with stimuli-responsive end-caps. 4.6. Synthesis of Polyphthalaldehydes by Cationic Polymerization. In addition to the anionic polymerization approach used in the work described above, cationic polymerization is another possible route for the synthesis of PPHA. However, the cationic polymerization method was much less explored until recently. In 2009, Ribitsch and co-workers (72) found that PPHA prepared usingaBF3OEt2 initiator could be isolated without end-capping, suggesting stability above its ceiling temperature. While the authors of this study specu- lated possible chain entanglements at high molecular weight as the reason for the unexpected stability, Moore and co-workers investigated this cationic polymerization in more detail to better understand the end-capping (73). They found that the cationic polymerization was much more rapid than the anionic polymerization, providing polymer within minutes, compared to hours for the anionic polymerization. In addition, while end-cap peaks were clearly visible in the NMR spectra of anionically synthesized PPHA, no end-cap peaks were observed in the spectra of cationically prepared PPHA of similar molecular weight. This led the authors to propose that the products of the cationic polymerization were cyclic species, a hypothesis that was confirmed by careful matrix- assisted laser desorption/inionization-time of flight mass spectrometry (MALDI- TOF MS) and SEC analysis of the polymers. Furthermore, they found that the SELF-IMMOLATIVE POLYMERS 21

Fig. 19. (a) The polymerizability of benzaldehyde derivatives increases with increasing Hammett values (reproduced with permission from reference (71). Copyright 2013 American Chemical Society.) (b) Reactions of halide-functionalized benzaldehydes incorporated into PPHA. (c) Reactions of aldehyde-functionalized benzaldehydes incorporated into PPHA. macrocyclization was reversible in nature, with the possibility to reopen the ring and expand or contract it in the presence of BF3OEt2, depending on the monomer concentration (Fig. 20a). While these cyclic polymers also do not contain stimuli- responsive end-caps, this intriguing behavior, resulting from polymerization near the ceiling temperature offers a new approach for the preparation of highly pure 22 SELF-IMMOLATIVE POLYMERS

Fig. 20. (a) Proposed mechanisms for the formation of cyclic PPHA by cationic polymerization and for its ring expansion-contraction. (Adapted with permission from reference (73). Copyright 2013 American Chemical Society.) (b) Schematic illustration of the synthesis of PPHA macrocycles and their scrambling. (Reproduced with permission from reference (74). Copyright 2013 American Chemical Society.) cyclic polymers, which are challenging to prepare by other processes, while at the same time providing dynamic properties. Based on this discovery, Moore and co-workers have further demonstrated that these cyclic polymers of OPA and its derivatives can ring open to depolymerize and exchange monomers to form new cyclic block copolymers and even random copolymers under the cationic polymerization conditions (Fig. 20b) (74). 4.7. Mechanically Triggered Depolymerization of Polyphthalaldehy- des. Moore, Boydston, and co-workers have also recently demonstrated the mechanically triggered depolymerization of PPHA and its subsequent repolymeriza- tion as a model of the continuous remodeling of biomaterials such as bone (75). –1 They prepared cyclic PPHA with Mns of 16.5, 58.2, and 254 kg mol by cationic polymerization using BF3OEt2, as well as a high molecular weight (Mn = 86.4 kg mol–1) poly(methyl acrylate) (PMA) control polymer. Mechanical scission was induced by pulsed ultrasound, resulted in depolymerization of the higher molecular weight PPHAs as demonstrated by SEC. Consistent with previous results on the ultrasound-induced depolymerization of polystyrene, which showed that a molecular weight threshold of 30 kg mol–1 was required for chain scission (76), the smallest PPHA did not undergo ultrasound-induced depolymerization. The PMA control sample underwent some degree of chain scission in response to SELF-IMMOLATIVE POLYMERS 23 ultrasound, but was not broken down to monomer units, demonstrating the fundamental difference between the low ceiling temperature, depolymerizable PPHA, and the PMA control. Molecular dynamics calculations as well as trapping studies suggested that PPHA was cleaved by a heterolytic scission mechanism, which is unusual in the context of mechanochemical bond scission. In addition, using an anionic polymerization initiated by n-butyllithium in THF at –78°C, it was possible to repolymerize the product monomer to regenerate polymer, demonstrating the potential application of these polymers in self-healing materials. In related work, Boydston and co-workers used PPHA to study the kinetics of the mechanochemical bond scission (77). Typically, polymers are cleaved into two fragments of similar molecular weight. If the fragments have molecular weights greater than the lower limit, these can undergo further scission, compli- cating the kinetic analysis. It was proposed that the rapid depolymerization of the PPHA daughter fragments would simplify the analysis. Indeed, through the –1 sonication of a single polydisperse sample of PPHA (Mn = 27 kg mol ;PDI= 1.5) and sectional analysis using 5 kg mol–1 ranges between 35 and 75 kg mol–1, the authors were able to elucidate a linear dependence of the rate constant for chain scission on molecular weight. They also observed a significant solvent dependence with the rate being significantly faster in DMF than in THF, likely due to the decreased cavitation efficiency of solvents with higher vapor pressure. However, some degradation of the OPA depolymerization product was observed in DMF whereas OPA was cleanly produced in THF. This would have important implications if the goal were to repolymerize the OPA in a self-healing or recon- figurable material. 4.8. Introduction of Polyglyoxylates as SIPs. Polyglyoxylates are another class of polyacetals with relatively low ceiling temperatures of less than 100°C in the solid state and much lower in solution (78,79). As such, they exhibit the tendency to rapidly depolymerize when not end-capped. Polyglyoxylates such as poly(methyl glyoxylate) (PMeG) (80,81), poly(ethyl glyoxylate) (PEtG) (78,79), and salts of poly(glyoxylic acid) (82,83) have been known for a few decades and have been investigated in potential applications such a biodegradable complex- ing agents and detergent builders (82,83) as well as for drug delivery (80,84). To prevent their spontaneous depolymerization in these applications, they were end-capped with vinyl ethers (82,83) or isocyanates (78–81,85) to afford acetal or carbamate end-capped polyglyoxylates, respectively. With these non-stimuli- responsive end-caps, the polyglyoxylates degrade via the cleavage of backbone acetal units, hydrolysis of pendant ester groups, and backbone depolymerization following scission (Fig. 21) (79,86,87). The initial degradation products are glyoxylic acid hydrate as well as the pendant alcohol. Glyoxylic acid is a metabolic intermediate in the glyoxylic acid cycle, a variation of the tricarboxylic acid cycle, occurring in plants, bacteria, and protists, which can convert glyoxylic acid to CO2. The degradation products of PEtG have been demonstrated to be nontoxic in invertebrate models and in plant ecotoxicity models (88), making polyglyoxylates of great interest as biodegradable polymers in general, but in particular for biomedical or agricultural applications where the potential toxicity of depolymerization products such as OPA (89) or quinone methides (50) would be problematic. In addition, the commercial availability of starting monomers makes these polymers particularly attractive relative to the various polycarbamates and related 24 SELF-IMMOLATIVE POLYMERS

Fig. 21. Mechanisms of degradation for PEtG without a stimuli-responsive end-cap.

derivatives, where the polymerization monomers typically require multiple steps of chemical synthesis. Recognizing the potential of polyglyoxylates as a new class of SIPs, Gillies and co-workers recently demonstrated that it was possible to introduce stimuli- responsive end-caps (90). In addition to the previously reported isocyanate end- caps (81), they found that it was possible to use chloroformates, enabling the introduction of a UV light-sensitive 6-nitroveratryl carbonate end-cap (Fig. 22). The polymerization was performed in CH2Cl2 at –20°C with catalytic NEt3.High monomer purity was critical for obtaining high molecular weight polymers in the range of 31–53 kg mol–1. Upon irradiation with UV light, PEtG depolymerized over several days in 9:1 acetonitrile:water forming ethyl glyoxylate hydrate. Thin ﬁlms of the material also degraded upon UV irradiation. In the absense of UV light or upon application of UV light to a control polymer without a UV-sensitive end-cap, the polymers were stable over the same time period. The approach was also extended to other polyglyoxylates including PMeG, poly(n-butyl glyoxylate), and poly(benzyl glyoxylate), whose glyoxylate monomers could be prepared from readily available starting materials such as maleic or fumaric acid. Random copolymers of these monomers with ethyl glyoxylate (EtG) were also prepared. This enabled tuning of the polymer glass transition temperature from –9°Cfor PEtG to 25°C for PMeG. Furthermore, using a multifunctional end-cap with a chloroformate as well as an alkyne, it was possible to prepare triblock copolymers with poly(ethylene glycol) (PEG) via a copper(I) assisted azide-alkyne cycloaddition reaction (Fig. 23). The resulting amphiphilic block copolymer self-assembled in aqueous solution to form micelles that depolymerized in minutes upon UV SELF-IMMOLATIVE POLYMERS 25

Fig. 22. Schematic of the synthesis of PEtG with different end-caps and the depolymerization of the 6-nitroveratryl carbonate end-capped polymer upon irradiation with UV light. irradiation. This work suggests the potential of these polyglyoxylates in applications such as drug delivery, involving encapsulation and rapid stimulus-triggered release. 4.9. Polyphthalaldehyde-Polyglyoxylate Copolymers. Kaitz and Moore have also recognized the value of using EtG in copolymers with OPA to improve the mechanical and thermal properties of PPHA (91). The authors noted that PPHA ﬁlms are highly brittle with poor mechanical toughness and a tendency to crack. They proposed that copolymerization with ethyl glyoxylate would enable tuning of the polymer Tg between that of the two homopolymers and lead to improved ﬁlm properties. Copolymerizations were performed cationically with BF3OEt2 as the catalyst at –78°C. At a 1:1 monomer feed ratio, it was found that nearly perfect alternating cyclic copolymers were formed. Tendencies toward 26 SELF-IMMOLATIVE POLYMERS

Fig. 23. Synthesis of a PEG-PEtG-PEG triblock copolymer.

alternating structures were also observed at other feed ratios as evidenced by 13C NMR and MALDI-TOF MS. This tendency was attributed to a combination of steric and electronic factors. As anticipated, the Tg of the copolymers could be tuned based on the monomer composition, ranging from 40°C for 38% OPA incorporation to 129°C for 63% OPA, in comparison with –20°C for pure PEtG in their study to a predicted Tg of 184°C for pure PPHA, a temperature higher than its decomposition temperature (Fig. 24). The decomposition temperature also depended on the copolymer composition, with those having lower OPA content being more stable. However, it was also noted that at high EtG content, this cationic polymerization approach led to lariat-shaped (cyclic polymer with pendant linear chain) polymer structures via backbiting, consistent with the group’s prior work on cationic homopolymerization of EtG (92), and that these copolymers were unstable due to the absence of end-cap. Overall, this work demonstrates the possibility to tune the properties of SIPs using combinations of OPA and glyoxylates.

5. Poly(benzyl ether)s

Phillips and co-workers developed end-capped self-immolative poly(benzyl ether)s with the aim of introducing a SIP that was easily prepared with lengths SELF-IMMOLATIVE POLYMERS 27

260

220

180

140

100 Temperature (°C) Temperature 60 Glass transition temperature Degradation temperature 20 Non-end-capped polymers –20 0 20 40 60 80 100 Percentage of OPA in copolymer

Fig. 24. Variation in the Tg and decomposition temperature of OPA-EtG copolymers as a function of polymer composition. (Reproduced with permission from reference (91). Copy- right 2013 American Chemical Society.)

Fig. 25. Chemical structure and depolymerization mechanism of a poly(benzyl ether). (Adapted with permission from reference (93). Copyright 2013 American Chemical Soci- ety.) up to hundreds or thousands of repeat units, was stabile to acid, base, and heat, was easily functionalized with an end-cap, exhibited solubility in common solvents, and underwent rapid depolymerization in a wide range of environments (93). Their approach built on the work of McGrath and co-workers on benzyl ether oligomers (11,25,94). They employed methyl substituents on the monomer to prevent its uncontrolled polymerization as well as a pendant phenyl group that would provide an enthalpic driving force for depolymerization through conjuga- tion with the quinone methide depolymerization product (Fig. 25). The polymerization was conducted anionically using isopropanol or methanol as an initiator and P2-t-Bu as a catalyst at –10°C for 1 h. Depending on the initiator to monomer ratio, polymerization time and temperature, and even purity of the monomer, the 28 SELF-IMMOLATIVE POLYMERS molecular weights of the polymers ranged from 3.6 to 484 kg mol–1. After polymerization, they reacted the polymers with a series of end-caps that were sensitive to ﬂuoride ions, UV light, and palladium(0). The depolymerization rate depended on the polarity of solvent. For example, depolymerization occurred in minutes in DMF and less than 2 days in THF, but required more than 1 week in toluene. Nevertheless, these depolymerization rates were still faster than many of the SIP backbones described above. In addition, compared to the previously investigated SIPs, this poly(benzyl ether) has better stability to acid, base, and heat, which may be useful in applications where poor stability and/or slow depolymerization prevent the use of other backbones.

6. Kinetics of Depolymerization

It is intuitive that longer SIPs should require longer depolymerization times than shorter SIPs of the same backbone, due to the requirement for more reactions to occur to completely break down the polymer backbone to small molecules. Recently, Gillies and co-workers have performed detailed kinetic studies using their previously reported polycarbamate backbone derived from 4-hydroxybenzyl alcohol and N,N-dimethylethylene diamine (Fig. 12a) to explicitly demonstrate this property (95). They prepared a series of monodisperse oligomers (monomer, dimer, tetramer, and octamer) by stepwise synthesis and used a design of experiments to optimize the conditions for preparation of two linear polymers with –1 –1 varying chain lengths (Mn = 5250 g mol with Đ of 1.47 and Mn = 13,600 g mol with Đ of 1.58). All of these molecules were prepared with Boc end-caps as the protonated amine terminus arising from cleavage of the end-cap with acid is stable to depolymerization until transferred to buffer, allowing end-cap cleavage to be decoupled from depolymerization. A kinetic model was developed in which cyclization was assumed to be slower than 1,6-elimination under the conditions of the study, and therefore the depolymerization could be described as a series of ﬁrst-order intramolecular reactions:

−→M −→M −→M −→M Pn Pn−1 ... P1 M(1)

kn kn−1 k2 k1

where Pi is the polymer chain of length i, M is the released monomer unit, and ki is the rate of intramolecular cyclization for a terminal cyclization spacer on a polymer chain of length i. It was assumed that the rate constant for intramolecular cyclization was independent of chain length, as this has been demonstrated to hold true for dendritic systems (10). Therefore, equation (1) was reduced to a set of linear differential equations:

d[Pn] =−k[Pn](2) dt SELF-IMMOLATIVE POLYMERS 29

Table 1. Kinetic Parameters for the Degradation of Monodisperse Oligomers

1 −1 a b Compound k × 10 (min ) t50 (min) Monomer 1.61 ± 0.37 4.5 ± 0.9 Dimer 1.48 ± 0.10 8.1 ± 0.9 Tetramer 1.60 ± 0.38 13.2 ± 1.7 Octamer 1.73 ± 0.34 23.9 ± 4.3

d [Pi] = k([Pi+1] − [Pi]) ∀i ≤ n − 1(3) dt where [Pi] is the concentration of polymer chains of length i and t is the time elapsed in the degradation process. Equations (2) and (3) were then reduced using an integrating factor to

i i−j −kt (kt) [Pn−i] = e [Pn−j ]0 (4) (i − j)! j=0

Released monomer was the most easily measured quantity during the depolymerization process. Although it is not a direct measure of polymer molecular weight, it is inversely related to molecular weight and could be used to measure polymer degradation according to equations (5) and (6):

n d [M] = k P (5) t i d i=1

= M(t) = M(t) D (t) n (6) M∞ i=1 i[Pi]0

In the case of the monodisperse oligomers, equations (2) and (3), and (5) can be al- gebraically solved and then substituted into equation (6) to provide the following equation describing the self-immolative depolymerization: ⎛ ⎞ n i j−1 M (t) ⎝ −kt −1 (kt) ⎠ D (t) = = [Pn]0 1 − e n (7) n[Pn] (j − 1)! 0 i=1 j=1

Oligomer depolymerization was studied in 3:2 pH 7.4 phosphate buffer:acetone at 37°Cby1H NMR spectroscopy, and the data were ﬁtted to equation (7) using nonlinear regression (Fig. 26a). It was found that indeed the rate constant for one monomer cyclization was independent of chain length within experimental error, and the time to 50% depolymerization (t50), as a measured of the depolymerization time, increased linearly with chain length (Table 1). 30 SELF-IMMOLATIVE POLYMERS

Fig. 26. (a)and(b): Degradation kinetics as measured by 1H NMR spectroscopy in 0.1 M phosphate buffer (D2O):acetone-d6 (3:2) at 37˚C. (a) Monomer (•), dimer (), tetramer (), octamer (). Solid lines correspond to the regressed fits of equation (7). (b) Degradation –1 –1 kinetics of polymers Mn = 5250 g mol with Đ of 1.47 ()andMn = 13,600 g mol with Đ of 1.58 (). Overlayed lines correspond to the self-immolative model fits for both polymers, (c) Mixed-mode degradation profile for the depolymerization of linear SIPs involving an initial zero-order domain followed by a gradual transition toward first-order behavior. (Reproduced with permission from reference (95). Copyright 2013 American Chemical So- ciety.)

Depolymerization of the polymers was studied under the same conditions as the oligomers, and their degradation profiles were fitted to the above depolymerization model using a nonlinear regression algorithm in which a numerical solution from equations (2) and (3), and (5) was applied to the SEC chromatograms as an indicator of the molecular weight distributions of the polymers (Fig. 26b). Again, good fits were obtained and the time to 50% depolymerization was pro- portional to chain length with the shorter polymer having a t50 of 17.4 min and the longer polymer having a t50 of 41.5 min; however, a limitation of this model is the requirement for a priori knowledge of the distribution of absolute polymer lengths, which is challenging to obtain and can only be approximated by SEC. As noted by McBride and Gillies (95), an analysis of the depolymerization kinetics of SIPs reveals an interesting mixed-mode phenomenon relatively unique to this class of polymers. In contrast to conventional biodegradable polymers that degrade by random backbone cleavage and typically display pseudo–first-order degradation kinetics, SIPs exhibit an initial pseudo–zero-order phase, followed by a transition to pseudo–first-order behavior during the course of depolymerization (Fig. 26c). This behavior arises because the concentration of polymer chains does not change until they are completely degraded to monomer units. Both SELF-IMMOLATIVE POLYMERS 31 experimental data and simulation studies indicate that this mixed-mode phenomenon is most apparent in monodisperse or low polydispersity samples, whereas as Đ is increased, the kinetics become more heavily weighted toward both short and long chains, resulting in an increased dispersity in times over which the transition from zero-order to first-order behavior occurs. For this reason, the depolymerization kinetics of SIPs may appear to be first order; however, the fitting of such data to first-order models is not strictly correct.

7. Conclusions and Prospects for Future Work

Although self-immolative linear polymers were only introduced less than a decade ago, there has been tremendous progress in their development and application. The field has benefitted significantly from prior work with self-immolative dendrimers and oligomers, which employed combinations of self-immolative spacers to provide degradation reaction cascades initiated by stimuli and which have undergone continued development alongside SIPs. It has also utilized the well- known concept of polymer ceiling temperature, with the major new development being the controlled introduction of stimuli-responsive end-caps to low ceiling temperature polymers. This has allowed unprecedented control over the depolymerization process. As described above, significant progress has been made in the development of synthetic approaches to SIPs. Careful monomer design has been employed to enable the preparation of polymeric materials instead of low molecular weight oligomers or monomer degradation products that could arise from competing elimination or cyclization reactions of the monomers. The principles of physi- cal organic chemistry, combined with synthetic tools, have been used to tune the rates of both the elimination and cyclization reactions, affording some control over the depolymerization rate. In the case of the polyacetals, both cationic and anionic polymerization approaches have been explored and optimized to afford high molar mass products over reasonable reaction times. It has also been shown both for PPHAs and for polyglyoxylates that comonomers can be incorporated, allowing the polymer properties such as Tg to be tuned and for the introduction of reactive sites. A wide array of end-caps has also been explored as triggering moieties for SIPs. These include small molecules susceptible to enzymatic cleavage, light, oxidizing or reducing agents, fluoride ions, and even heat. A unique advan- tage of SIPs relative to conventional stimuli-responsive polymers is that a single polymer backbone can be imparted with responsiveness to numerous stimuli sim- ply by changing the end-cap. Another unique attribute of SIPs relative to other classes of degradable or stimuli-responsive polymers is their degradation kinetics, which displays an interesting combination of pseudo–zero-order and first- order behavior. It has now been explicitly demonstrated that the time required for depolymerization depends on the degree of polymerization in a predictable manner. The most important feature of SIPs is their amplified response to stimuli, which results from the fact that a single end-cap cleavage results in a cascade of reactions. This feature promises great potential for SIPs in applications, and significant progress has already been made in demonstrating the potential of these 32 SELF-IMMOLATIVE POLYMERS polymers in a number of fields. For example, SIPs have been explored as small molecule and enzymatic sensors both in solution and in paper-based devices, where their amplified response affords increased sensitivity. They have also been incorporated into amphiphilic blocks, enabling the preparation of nanoscale assemblies such as micelles and vesicles, where their depolymerization in response to stimuli results in degradation of the assemblies and the release of cargo such as drug molecules. They have also been explored as shell wall materials for microcapsules that exhibit triggered release, with the potential to provide self-healing capabilities. Despite the tremendous progress in developing the chemistry and proof- of-principle applications of SIPs, there is still significant work required to expand the array of SIP backbones and applications and to translate these excit- ing findings to the market. The number of different linear SIPs backbones is still relatively small. Considering the vast array of potential small molecules and functional groups available to chemists, many potential classes of SIPs remain unexplored at this time. Therefore, it is anticipated that new SIP backbones will continue to be developed. With this goal in mind, consideration must also be given to the synthetic accessibility of the required monomers and the efficiencies of the overall synthetic routes for preparing SIPs. Thus far, only the polyacetal class of SIPs can be prepared from inexpensive and readily available monomers. In addition, aside from studies on PEtG (88) and the known degradation products such as OPA (89) and quinone methides (50), there have so far been very few investigations regarding the potential toxicities of the SIP degradation products. This will be particularly important for biomedical applications such as drug delivery and in vivo sensing, but should also be a general consideration for any application as the degradation products would eventually be released into the environment through various mechanisms. Finally, it will be necessary to follow up promising proof-of-principle results in applications such as sensing and encapsulation with further studies and device optimization to develop products that will be highly competitive from both a cost and performance standpoint for translation to the market.

BIBLIOGRAPHY

1. A. Chunder, K. Etcheverry, G. Londe, H. J. Cho, and L. Zhai, Colloids Surf. A 333, 187–193 (2009). 2. Z. Ge and S. Liu, Chem. Soc. Rev. 42, 7289–7325 (2013). 3. E. S. Lee, K. Na, and Y. H. Bae, Nano Lett. 5, 325–329 (2005). 4. Y. Lee, T. Ishii, H. Cabral, H. J. Kim, J. H. Seo, N. Nishiyama, H. Oshima, K. Osada, and K. Kataoka, Angew. Chem., Int. Ed. 48, 5309–5312 (2009). 5. E. R. Gillies and J. M. J. Frechet,´ Chem. Commun. 1640–1641 (2003). 6. M. J. Hefferman and N. Murthy, Bioconjugate Chem. 16, 1340–1342 (2005). 7. C. Lin, Z. Y. Zhong, M. C. Lok, X. L. Jiang, W. E. Hennink, J. Feijan, and J. F. J. Engbersen, Bioconjugate Chem. 18, 138–145 (2007). 8. H. Zhao, E. S. Sterner, E. B. Coughlin, and P. Theato, Macromolecules 45, 1723–1736 (2012). 9. F. M. H. deGroot, C. Albrecht, R. Koekkoek, P. H. Beusker, and H. W. Scheeren, Angew. Chem., Int. Ed. 42, 4490–4494 (2003). SELF-IMMOLATIVE POLYMERS 33

10. R. J. Amir, N. Pessah, M. Shamis, and D. Shabat, Angew. Chem., Int. Ed. 42, 4494– 4499 (2003). 11. M. L. Szalai, R. M. Kevwitch, and D. V. McGrath, J. Am. Chem. Soc. 125, 15688–15689 (2003). 12. P. L. Carl, P. K. Chakravarty, and J. A. Katzenellenbogen, J. Med. Chem. 24, 479–480 (1981). 13. F. Kratz, I. A. Muller,¨ C. Ryppa, and A. Warnecke, ChemMedChem 3, 20–53 (2008). 14. M. Wakselman, Nouv.J.Chem.7, 439–447 (1983). 15. D. Shan, M. G. Nicolaou, R. T. Borchardt, and B. Wang, J. Pharm. Sci. 86, 765–767 (1997). 16. M. Shamis, H. N. Lode, and D. Shabat, J. Am. Chem. Soc. 126, 1726–1731. (2004). 17. R. Erez, E. Segal, K. Miller, R. Satchi-Fainaro, and D. Shabat, Bioorg. Med. Chem. 17, 4327–4335 (2009). 18. E. Sella and D. Shabat, Chem. Commun. 2008, 5701–5703 (2008). 19. E. Sella and D. Shabat, J. Am. Chem. Soc. 131, 9934–9936 (2009). 20. E. Sella, R. Weinstain, R. Erez, N. Z. Burns, P. S. Baran, and D. Shabat, Chem. Com- mun. 46, 6575–6577 (2010). 21. Y. Meyer, J. A. Richard, M. Massonneau, P. Y. Renard, and A. Romieu, Org. Lett. 10, 1517–1520 (2008). 22.F.M.H.deGroot,W.J.Loos,R.Koekkoek,L.W.A.vanBerkom,G.F.Busscher,A. E. Seelen, C. Albrecht, P. deBrujin, and H. W. Scheeren, J. Org. Chem. 66, 8815–8830 (2001). 23. R. Weinstain, E. Segal, R. Satchi-Fainaro, and D. Shabat, Chem. Commun. 46, 553– 555 (2010). 24. A. Warnecke and F. Kratz, J. Org. Chem. 73, 1546–1552 (2008). 25. S. Li, M. L. Szalai, R. M. Kevwitch, and D. V. McGrath, J. Am. Chem. Soc. 10516–10517 (2003). 26. O. Vogl, J.Polym.Sci.,A:Polym.Chem.38, 2293–2299 (2000). 27. W. M. D. Bryant and J. B. Thompson, J. Polym. Sci. A-1 9, 2523 (1971). 28. D. J. Darensbourg and S. H. Wei, Macromolecules 45, 5916–5922 (2012). 29. D. J. Darensbourg, S.-H. Wei, A. D. Yeung, and W. C. Ellis, Macromolecules 46, 5850– 5855 (2013). 30. D. J. Darensbourg, S. H. Wei, and S. J. Wilson, Macromolecules 46, 3228–3233 (2013). 31. D. J. Darensbourg, A. D. Yeung, and S. H. Wei, Green Chem. 15, 1578–1583 (2013). 32. A. D. Wong, M. A. DeWit, and E. R. Gillies, Adv. Drug Delivery Rev. 64, 1031–1045 (2012). 33. G. I. Peterson, M. B. Larsen, and A. J. Boydston, Macromolecules 45, 7317–7328 (2012). 34. S. T. Phillips and A. M. DiLauro, ACS Macro Lett. 3, 298–304 (2014). 35. S. T. Phillips, J. S. Robbins, A. M. DiLauro, and M. G. Olah, J. Appl. Polym. Sci. 131, 40992–41003 (2014). 36. A. Sagi, R. Weinstain, N. Karton, and D. Shabat, J. Am. Chem. Soc. 130, 5434–5435 (2008). 37. R. Weinstain, A. Sagi, N. Karton, and D. Shabat, Chem. Eur. J. 14, 6857–6861 (2008). 38. R. Weinstain, P. S. Baran, and D. Shabat, Bioconjugate Chem. 20, 1783–1791 (2009). 39. G. G. Lewis, J. S. Robbins, and S. T. Phillips, Macromolecules 46, 5177–5183 (2013). 40. G. G. Lewis, M. J. DiTucci, and S. T. Phillips, Angew. Chem., Int. Ed. 51, 12707–12710 (2012). 41. G. G. Lewis, J. S. Robbins, and S. T. Phillips, Anal. Chem. 85, 10432–10439 (2013). 42. G. G. Lewis, J. S. Robbins, and S. T. Phillips, Chem. Commun. 50, 5352–5354 (2014). 43. J. S. Robbins, K. M. Schmid, and S. T. Phillips, J. Org. Chem. 78, 3159–3169 (2013). 44.G.I.Peterson,D.C.Church,N.A.Y.,andA.J.Boydston,Polymer 09, 048 (2014). 34 SELF-IMMOLATIVE POLYMERS

45. B. Zeng, J. Huang, M. Wright, and S. King, Med. Chem. Lett. 14, 5565–5568 (2004). 46. A. P. Esser-Kahn, N. R. Sottos, S. R. White, and J. S. Moore, J. Am. Chem. Soc. 132, 10266–10268 (2010). 47. G. Liu, X. Wang, J. Hu, G. Zhang, and S. Liu, J. Am. Chem. Soc. 136, 7492–7497 (2014). 48. D. E. Discher and A. Eisenberg, Science 297, 967–973 (2002). 49. M. Brasch, I. K. Voets, M. S. T. Koay, and J. J. L. M. Cornelissen, Faraday Discuss. 166, 47–57 (2013). 50. T. Monks and D. Jones, Drug Metab. 3, 425–438 (2002). 51. M. A. DeWit and E. R. Gillies, J. Am. Chem. Soc. 131, 18327–18334 (2009). 52. M. A. DeWit, A. Nazemi, S. Karamdoust, and E. R. Gillies, ACS Symp. Ser. 1066, 9–21 (2011). 53. C. deGracia Lux, C. L. McFearin, S. Joshi-Barr, J. Sankaranarayanan, N. Fomina, and A. Almutairi, ACS Macro. Lett. 1, 922–926 (2012). 54. K. M. Schmid, L. Jensen, and S. T. Phillips, J. Org. Chem. 77, 4363–4374 (2012). 55. M. A. DeWit and E. R. Gillies, Org. Biomol. Chem. 9, 1846–1854 (2011). 56. E. K. Y. Chen, R. A. McBride, and E. R. Gillies, Macromolecules 45, 7364–7374 (2012). 57. M. A. Dewit, A. Beaton, and E. R. Gillies, J. Polym. Sci., A: Polym. Chem. 48, 3977– 3985 (2010). 58. P. Kuppusamy, H. Li, G. Ilangovan, A. J. Cardounel, J. L. Zweier, K. Yamada, M. C. Krishna, and J. B. Mitchell, Cancer Res. 62, 307–312 (2002). 59.G.Wu,Y.Z.Fang,S.Yang,J.R.Lupton,andN.D.Turner,J. Nutr. 134, 489–492 (2004). 60. P. Kuppusamy, M. Afeworki, R. A. Shankar, D. Cofﬁn, M. C. Krishna, S. M. Hahn, J. B. Mitchell, and J. L. Zweier, Cancer Res. 58, 1562–1568 (1998). 61. C. Aso, S. Tagami, and T. Kunitake, J. Polym. Sci., A: Polym. Chem. 7, 497–511 (1969). 62.C.G.Wilson,H.Ito,J.M.J.Frechet,T.G.Tessier,andF.M.Houlihan,J. Electrochem. Soc. 133, 181–187 (1986). 63. S. A. MacDonald, C. G. Willson, and J. M. J. Frechet, Acc. Chem. Res. 27, 151–158 (1994). 64. H. Ito and C. G. Willson, Polym. Eng. Sci. 23, 1012–1018 (1983). 65. W. Seo and S. T. Phillips, J. Am. Chem. Soc. 132, 9234–9235 (2010). 66. O. Coulembier, A. Knoll, D. Pires, B. Gotsmann, U. Duerig, J. Frommer, R. D. Miller, P. Dubois, and J. L. Hedrick, Macromolecules 43, 572–574 (2010). 67. A. M. DiLauro, J. S. Robbins, and S. T. Phillips, Macromolecules 46, 2963–2968 (2013). 68.H.Zhang,K.Yeung,J.S.Robbins,R.A.Pavlick,M.Wu,R.Liu,A.Sen,andS.T. Phillips, Angew. Chem., Int. Ed. 51, 2400–2404 (2012). 69. A. M. DiLauro, H. Zhang, M. S. Baker, F. Wong, A. Sen, and S. T. Phillips, Macro- molecules 46, 7257–7265 (2013). 70. A. M. DiLauro, A. Abbaspourrad, D. A. Weitz, and S. T. Phillips, Macromolecules 46, 3309–3313 (2013). 71. J. A. Kaitz and J. S. Moore, Macromolecules 46, 608–612 (2013). 72. S. Kostler,¨ B. Zechner, B. Trathnigg, H. Fasl, W. Kern, and V. Ribitsch, J. Polym. Sci., Polym. Chem. 47, 1499–1509 (2009). 73. J. A. Kaitz, C. E. Diesendruck, and J. S. Moore, J. Am. Chem. Soc. 135, 12755–12761 (2013). 74. J. A. Kaitz, C. E. Diesendruck, and J. S. Moore, Macromolecules 46, 8121–8128 (2013). 75. C. E. Diesendruck, G. I. Peterson, H. J. Kulik, J. A. Kaitz, B. D. Mar, P. A. May, S. R. White, T. J. Martinez, A. J. Boydston, and J. S. Moore, Nat. Chem. 6, 623–628 (2014). 76. T. Q. Nguyen, Q. Z. Liang, and H.-H. Kausch, Polymer 38, 3783–3793 (1997). 77. G. I. Peterson and A. J. Boydston, Macromol. Rapid Comm. 35, 1611–1614 (2014). 78. F. Burel, L. Rossignol, P. Pontvianne, J. Hartman, N. Couesnon, and C. Bunel, e- Polymers 3, 407–418 (2003). SELF-IMMOLATIVE POLYMERS 35

79. B. Belloncle, F. Burel, H. Oulyadi, and C. Bunel, Polym. Degrad. Stab. 93, 1151–1157 (2008). 80. C.-H. Brachais, R. Duclos, C. Vaugelade, J. Huguet, M. L. Capelle-Hue, and C. Bunel, Int. J. Pharm. 169, 23–31 (1998). 81. C. H. Brachais, J. Huguet, and C. Bunel, Polymer 38, 4959–4964 (1997). 82. U.S. Pat. 4144226 (to Monsanto Co) (1979), M. M. Crutchﬁeld, V. D. Papanu, and C. B. Warren. 83. U.S. Pat. 4140676 (to Monsanto Co) (1979), M. M. Crutchﬁeld and C. J. Upton. 84. J. K. Kim, V. K. Garripelli, U. H. Jeong, J. S. Park, M. A. Repka, and S. Jo, Int. J. Pharm. 401, 79–86 (2010). 85. B. Belloncle, F. Burel, and C. Bunel, Polym. Prepr. 48, 633–634 (2007). 86. L. Brachais, C.-H. Brachais, J. Huguet, and C. Bunel, Polym. Degrad. Stab. 64, 243– 249 (1999). 87. C. H. Brachais, J. Huguet, C. Bunel, and L. Brachais, Polymer 39, 883–890 (1998). 88. B. Belloncle, C. Bunel, L. Menu-Bouaouiche, O. Lesouhaitier, and F. Burel, J. Polym. Environ. 20, 726–731 (2012). 89. S. Anderson, C. Umbright, R. Sellamuthu, K. Fluharty, M. Kashon, J. Franko, L. Jack- son, V. Johnson, and P. Joseph, Toxicol Sci. 115, 435–443 (2010). 90. B. Fan, J. F. Trant, A. D. Wong, and E. R. Gillies, J. Am. Chem. Soc. 136, 10116–10123 (2014). 91. J. A. Kaitz and J. S. Moore, Macromolecules 47, 5509–5513 (2014). 92. J. A. Kaitz, C. E. Diesendruck, and J. S. Moore, Macromolecules 47, 3603–3607 (2014). 93. M. G. Olah, J. S. Robbins, M. S. Baker, and S. T. Phillips, Macromolecules 46, 5924– 5928 (2013). 94. R. M. Kevwitch, C. S. Shanahana, and D. V. McGrath, New J. Chem. 36, 492–505 (2012). 95. R. A. McBride and E. R. Gillies, Macromolecules 46, 5157–5166 (2013).

BO FAN ELIZABETH R. GILLIES The University of Western Ontario London, Canada