Controlled organocatalyzed d,l-lactide ring-opening polymerizations: synthesis of low molecular weight oligomers

A systematic approach to the synthesis of organocatalyzed oligo(d,l-lactide) demonstrates that choice of initiator, catalytic ratio, and reaction time yields well-controlled oligomers. Ring-opening polymerization of d,l-lactide with the initiator α-methyl propargyl alcohol, a secondary alcohol, used in excess of 4-dimethylaminopyridine catalyst mitigates cyclicization, transesterification, and catalyst-initiated side reactions. This approach enables the design of uniform lactide oligomers for controlled release applications, such as delivery systems for drugs, prodrugs, and molecular sensors.

Poly(lactide) (PLA) is a versatile polymer with properties suitable for a range of controlled release applications. Molecular weight and copolymer composition control degradation and drug release, with degradation yielding biocompatible lactic acid products. PLA and copolymers of lactide and glycolide (PLGA) have been used pre-clinically and clinically for release of drugs, 1-3 prodrugs, 4-6 and molecular sensors. [7][8][9] Highly reproducible PLA chemistries are particularly important to ensure control over the release of drugs with narrow therapeutic windows and to enhance efficacy by reducing dependence on patient compliance. [10][11][12][13] Stannous octoate is commonly employed as an organometallic catalyst of lactide ring-opening polymerization (ROP). 14,15 However, tin catalysts, which are challenging to fully remove during purication, can result in toxicity. 16 Alternatives such as strongly basic amine organocatalysts are favored, particularly 4-dimethylaminopyridine (DMAP), which was pioneered by Nederberg et al. 17 DMAPmediated ROP is used for one-pot PLA polymerizations and conjugations, 18 diblock or triblock copolymerizations with lactide as a rst or second block, 19-21 graing lactide to cellulose polymer bers, 22 and synthesizing star-shaped/cross-linked PLA networks. 23 PLA has also been exploited to tether and release drugs from linear polymers and hydrogel depots. 24 Drug tethers are typically low molecular weight oligo(lactide) composed of fewer than seven lactide repeat units to avoid crystallinity and associated challenges with solubility and control over degradation rates. 25 Self-catalysis, or direct polycondensation, of lactic acid at increased temperature and reduced pressure for an extended time yields low-molecular weight (800-3200 Da) oligo(lactide), in contrast to organocatalysts that are commonly employed for high molecular weight PLA synthesis. However, extensive setup and environmental control decrease the accessibility of these reactions. 26,27 Furthermore, with neither an initiator nor a catalyst used in polycondensation, mixtures of ahydroxy and u-carboxy PLA are formed, limiting the versatility of post-polymerization drug functionalization. 28 Alternatively, click chemistry 29 may be exploited to control subsequent modication of oligo(lactide) by employing a 'clickable' alcohol to initiate oligo(lactide) synthesis. 30 Here, we explored bulk ROP of D,L-lactide by propargyl alcohol initiator and DMAP catalyst to synthesize low molecular weight oligo(lactide) linkers. To enable subsequent click reactions and to mitigate crystallinity, propargyl alcohol (PA) and D,L-lactide (L) were used. 27 Although not studied here, similar approaches have shown products do not epimerize, and we expect oligomers to be atactic and amorphous. 17 To investigate the molecular weight and polydispersity of oligo(lactide), an initial polymerization was designed similar to that of Nederberg et al. 17 and Coulembier and Dubois 18 using a PA : L : DMAP ratio of 1 : 20 : 4 (Scheme 1; see Table S1, ESI †). The neat polymerization was stirred at 130 C under a nitrogen environment. Aer 5, 10, 15, 30, and 60 minutes, reaction vials were opened to atmosphere and cooled before dissolution in dichloromethane (DCM) and precipitation in hexanes. 1 H-NMR spectroscopy identied successful synthesis of PA-functional oligo(D,L-lactide) (PA-ODLA) with 99% conversion of D,L-lactide aer only 5 minutes of polymerization ( Fig. 1a; see Fig. S3, ESI †). Integration of peaks C, E, and F indicated linkers were 19 lactide units, or an average M n of 2825 Da. However, matrix assisted laser desorption ionization time of ight mass spectrometry suggested M n was 752 Da. As M n determined by NMR was based on average end-group analysis and assumed PA-initiated oligo(lactide), and M n determined by MALDI analysis represented all species present, these data indicated that not all oligo(lactide) chains were initiated by PA. Rather, distinct series of peaks periodically separated by 144 Da, the M n of lactide, were formed during polymerization (a, b, g, d, 3, and z; Fig. 1b; see Table S2, ESI †). Peaks 72 Da less than these peaks were also identied (a 0z 0 ). Additional reactions were conducted to identify the formed products: one with the initiator a-methyl propargyl alcohol (aMPA; see Fig. S4, ESI †), and another with only D,L-lactide and DMAP but no initiator (see Fig. S9, ESI †).
The desired product, PA-ODLA "a" (Fig. 1a), was formed via DMAP base activation of PA, which initiated ROP (see Fig. S5, ESI †). Base activation of the alcohol was conrmed similar to previous reports 20,31 using 1 H-NMR of PA mixed with DMAP in CDCl 3 to identify ppm shis for hydroxyl groups (see Fig. S6, ESI †). Peak a 0 , 72 Da less than a, was the result of trans-esterication of PA-ODLA (see Fig. S7, ESI †), which is a common side-reaction during ROP of lactide. 18,20 Two undesired products, peaks z and z 0 , were cyclic PA-ODLA with and without transesterication, respectively. Cyclicization increases over time during ROP of lactide [32][33][34] and is undesired because hydroxyl end groups are not available for subsequent conjugation. These peaks were due to radical-mediated dehydration with hydroxyl end group participation, 35 as the addition of hydroquinone (HQ), a radical scavenger, eliminated peaks z and z 0 (see Fig. S8, ESI †). However, a peak appeared that corresponded to HQ-DMAP-ODLA and lacked alkyne functionalities for subsequent click reactions.
Another undesired product, DMAP-ODLA, or peak g, formed via DMAP nucleophilic attack of lactide (see Fig. S9, ESI †). Peak g 0 , separated by 72 Da from peak g, resulted from trans-esterication of DMAP-ODLA. By conducting reactions of only DMAP and lactide, only PA and lactide, and only aMPA and lactide, it was conrmed that DMAP can both initiate and catalyze lactide ROP, but PA and aMPA cannot (see Fig. S10, ESI †). DMAP-ODLA formed due to PA : DMAP ratios less than 1, as previously described, 18 and is undesired because 'clickable' propargyl end groups are not present. Finally, peaks b, d, and 3 are likely ion fragments, as these peaks were only present when using MALDI in linear, but not reector, mode (see Fig. S11, ESI †).
To increase the amount of oligo(lactide) with alkyne and hydroxyl functionalities, side reactions were systematically addressed. It was noted that transesterication increased with time of polymerization and was greater for PA than for aMPA, as less nucleophilic secondary alcohols are unable to participate in transesterication reactions. 30 Thus, reactions were conducted at 130 C for 5 minutes using aMPA as the initiator to optimize desired product (aMPA-ODLA, peak a).
Ratios of aMPA : L : DMAP were investigated to mitigate undesired DMAP-initiated and cyclic ODLA (Table 1). Holding aMPA : L constant and increasing aMPA : DMAP increased the intensity of cyclic aMPA-ODLA z relative to DMAP-ODLA g, but aMPA-ODLA a became negligible ( Fig. 2a;     Similar behavior was identied when holding L : DMAP constant and increasing aMPA : L ( Fig. 2c; see Fig. S14, ESI †). These results suggest that aMPA : L controls cyclicization, as the distribution of z is similar among reactions in Fig. 2a, and the ratio of z : a decreases between ratios of 1 : 10 and 1 : 2 in Fig. 2b. aMPA : DMAP controls DMAP-ODLA generation, as the distribution of g is similar within Fig. 2b and smallest at a ratio of 2 : 1. Finally, lower ratios of L : DMAP appear to increase a, with 2 : 1 yielding the greatest amount of desired product a. Altogether, higher aMPA : L, higher aMPA : DMAP, and lower L : DMAP ratios yield higher levels of a relative to g and z.
Increasing the stoichiometry of lactide to initiator is commonly exploited to form linkers of various lengths to achieve differential release rates. 24 To explore the utility of this approach using optimized reaction conditions, aMPA : L ratios of 1 : 2, 1 : 5, and 1 : 10 were investigated using 1 mol% DMAP ( Table 2, see Fig. S15-S17, ESI †). With an aMPA : DMAP ratio of 33 or greater, g was absent in all reactions. The polymerization with the highest aMPA : L ratio, 1 : 2 : 0.03, had the greatest overall a : z ratio. Interestingly, the peak h appeared, corresponding to aMPA-ODLA with a K + ion, in contrast to a with a DMAP-H + ion adduct. There was a transition from h to a over time, and h was most abundant in 1 : 2 : 0.03 reactions. This was explored further with polymerization times of 4 minutes and less (see Fig. S18, ESI †). One minute reactions showed negligible conversion of lactide monomer. Aer 2 minutes, conversion was 68% and the major product was h. By 4 minutes, a and z exceeded h, suggesting that shorter reaction times are necessary to isolate linear aMPA-ODLA without DMAP adducts.

Conclusions
This work presents a systematic design of narrowly dispersed, low molecular weight oligo(lactide) formed without metal catalysts that are difficult to completely remove during purication. Before optimization, observed MALDI M n disagreed with M n as determined by NMR, suggesting transesterication   reactions were present. Previous studies investigated the singular effects of polymerization temperature, polymerization time, monomer : initiator ratio, chemical initiator, and reagent degassing time on PLA transesterication. 36 This study identied the combined effects of choice of initiator, initiator : catalyst ratio, initiator : monomer ratio, and polymerization time on molecular weight, polydispersity, and relative amounts of desired and undesired products. Temperature was not changed from 130 C due to the high melting point of D,L-lactide (122 C). A 90 second reaction of 1 : 2 : 0.03 aMPA : L : DMAP formed oligo(lactide) with alkyne and hydroxyl functionalities available for subsequent click conjugations. Further optimization may be required to increase the reaction efficiency, as theoretical M n does not match M n determined by MALDI. Although studies herein investigated ROP of lactide by DMAP, the approach can be employed for similar systems (e.g. catalysis by stannous octoate or ROP of glycolide). Similarly, copper-catalyzed azide-alkyne cycloaddition (CuAAC) may be used in place of Huisgen 1,3-dipolar cycloaddition for applications in which residual copper contamination is acceptable.

Conflicts of interest
There are no conicts to declare.