Characterization of commercial PLGAs by NMR spectroscopy

Poly(lactic-co-glycolic acid) (PLGA) is among the most common of biodegradable polymers studied in various biomedical applications such as drug delivery and tissue engineering. To facilitate the understanding of the often overlooked impact of PLGA microstructure on important factors affecting PLGA performance, we measured four key parameters of 17 commonly used commercial PLGA polymers (Expansorb®, Resomer®, Purasorb®, Lactel®, and Wako®) by NMR spectroscopy. 1HNMR and 13CNMR spectra were used to determine lactic to glycolic ratio (L/G ratio), polymer end-capping, glycolic blockiness (Rc), and average glycolic and lactic block lengths (LG and LL). In PLGAs with a labeled L/G ratio of 50/50 and acid end-capping, the actual lactic content slightly decreased as molecular weight increased in both Expansorb® and Resomer®. Whether or not acid- or ester-, termination of these PLGAs was confirmed to be consistent with their brand labels. Moreover, in the ester end-capped 75/25 L/G ratio group, the blockiness value (Rc) of Resomer® RG 756S (Rc: 1.7) was highest in its group; whereas for the 50/50 acid end-capped group, Expansorb® DLG 50-2A (Rc: 1.9) displayed notably higher values than their counterparts. Expansorb® 50-2E (LL: 2.5, LG: 2.6) and Resomer® RG 502 (LL: 2.6, LG: 2.5) showed the lowest block lengths, suggesting they may undergo a steadier hydrolytic process compared to random, heterogeneously distributed PLGA.


Introduction
Polymeric biomaterials are widely used in commercial biomedical products in the past several decades [1]. Poly(lactic-co-glycolic acid) or poly(lactide-co-glycolide) (PLGA) is a thermoplastic co-polyester comprised of various ratios of its monomers and is hydrolyzed in vivo into non-toxic lactic and glycolic acid, that are metabolized in the tricarboxylic acid cycle and eliminated via carbon dioxide and water [2,3]. Microscopic PLGA properties such as L/G ratio, molecular weight (MW), and end-capping are primary polymer factors impacting the subsequent behaviors of PLGA formulations such as their degradation and drug release behavior [4]. Beyond these polymer properties, there are manufacturing and formulation variables such as formulation processes, size/geometry, presence of excipients, and the drug itself that all serve a considerable role in product performance, but the scope of this research is to investigate fundamental polymer properties [5,6]. The PLGA microstructures such as monomer sequence distribution along the polymer chain, described by the block length and blockiness, are often overlooked and rarely reported, though it is expected to have an effect on the hydrolytic degradation profile. The blockiness of PLGA measures the relative occurrence of glycolyl-glycolyl compared to glycolyl-lactyl linkages, while the block length measures the average length of the glycolic and lactic linkages in one segment [7]. The increased reactivity of glycolic monomers has been previously studied, and Vey et al. determined that the glycolic unit (G-G bond) consistently hydrolyzes 1.3 times faster than the lactic unit across different L/G ratio polymer films submerged in phosphate buffer [8][9][10]. Due to the increased reactivity of the glycolic-glycolic linkage, the increase in blockiness or block length will lead to an accelerated hydrolysis of PLGA. It was recently reported that sequenced PLGAs with a "true" alternating microstructure [11] exhibited a slower swelling, more gradual loss of molecular weight, and a longer preservation of morphology compared to the more "blocky" counterparts synthesized by ring-opening polymerization (ROP).
As a non-toxic biomaterial commonly used in many FDA-approved products, various commercial PLGA polymers across different brands and with different features (MW, L/G ratio, and end-capping) have been widely investigated. However, for many of these commercial PLGA polymers, properties such as block length and blockiness remain unclear and have yet to be systematically analyzed. Hence, in this work, using NMR spectroscopy, we investigated 17 commonly-used PLGAs separated into 4 different groups of similar composition, and obtained four critical chemical properties, namely, L/G ratio [12], end-capping [12], blockiness, and block length to compare [7]. We compared different polymer portfolios of 5 manufacturers, Expansorb®, Resomer®, Purasorb®, Lactel®, and Wako®. Not all manufacturers had the same breadth of polymers available; e.g., there are not two polymers from Expansorb® to compare with Lactel®; thus, not all brands were compared in each group. The potential impact of these data on physicochemical properties such as degradation is discussed and will be investigated directly in future publications and depth analysis of erosion behavior and drug release properties of these polymers. Hence, these results could help bridge the gap between the parameters obtained from PLGA microstructures and subsequent performance, providing guidance for further biomedical and drug delivery applications.

Nuclear magnetic resonance spectroscopy
NMR scanning was performed using a Varian vnmr-500 MHz (11.7 Tesla) Premium Shielded NMR spectrometer (USA) running Vnmrj software (the NMR Facility of the University of Michigan Chemistry Department, USA) for 1 HNMR and 13 CNMR. NMR spectra are available in the Supporting Information.

Lactic/glycolic ratio calculation
Each PLGA sample was weighed (~5 mg) and dissolved in deuterated chloroform (CDCl 3 ) (0.5 mL), then pipetted into an NMR tube. The spectrum of every sample was collected by 1 HNMR spectroscopy (16 scans, 0.5-s relaxation delay, and 45-degree pulse angle). L/G ratios were determined by comparing proton intensities at chemical shifts 5.2 ppm and 4.8 ppm [12]. The peak at 5.2 ppm represents a single proton of the lactic unit, while the peak of 4.8 ppm represents two protons of the glycolic unit. Hence, the mole fractions of lactic (m L ) and glycolic (m G ) units were calculated from peak integrations (p) of each component, as shown below:

End-capping analysis
Each PLGA sample was weighed (~15 mg), dissolved in deuterated chloroform (CDCl 3 ) (0.75 mL), and then the resulting polymer solutions pipetted into an NMR tube. The endcapping of PLGA polymers were confirmed by 13 CNMR. The relative molecular weight of the ester end-cap is smaller than the PLGA polymer; thus, the signal collection should be performed with the maximized signal-to-noise ratio. A Z-restored spin-echo pulse sequence was used with a 30-degree observation pulse, a 3-s inter-pulse delay, and a 0.55-s data acquisition time and a total of 12,000 scans were acquired over 12.5 h [13]. Generally, PLGA polymers have two types of end caps: ester and acid end caps. In the 13 C NMR spectrum, methyl units at the end of the alkyl chain appear very far upfield at a chemical shift of 14 ppm, which was used to confirm an ester end-capping of PLGA polymer. Acid end-capping was confirmed by the absence of this esterspecific peak at 14 ppm in their NMR spectrum [12].

Blockiness analysis
The blockiness was determined by high-resolution 13 CNMR [12]. These spectra were collected from 50 mg/mL polymer solutions (25 mg) in CDCl 3 (0.5 mL) and analyzed for the different monomer sequence distributions. This was done by using a pulse sequence without NOE enhancement, and employing a 30-degree 13 C observation pulse, a 2.0-s interpulse delay, a 4.6-s relaxation delay, and a 0.4-s acquisition time. Total acquisition time was approximately 4 h and 50 min to give an optimal signal-to-noise ratio.
Blockiness represents the heterogeneity of PLGA and is determined by the glycolic unit sequence distribution according to the glycolic carbonyl group located in the chemical shift 166-167 ppm in high-resolution 13 CNMR spectrum. The upfield G-G peak (I G-G ) represents the glycolyl-glycolyl carbonyl; while the downfield G-L peak (I G-L ) represents a glycolyl-lactyl carbonyl. Blockiness values can be calculated according to two methods. One is calculated by the definition of Skidmore et al. [14], in which Rc value is obtained by the ratio of I G-G to I G-L , while the other is the reverse, Rcms (the ratio of I G-L to I G-G ) by the definition of Hausberger and DeLuca [4] representing the ratio of the two carbonyl peaks. Here, we use the Rc value to describe blockiness.

Block length analysis
The block lengths were determined by high-resolution 13 CNMR [15]. Each PLGA sample was weighed (20 mg) and dissolved in a mixture of hexafluoro-2-propanol (10 mg/mL) for enhanced solubility and deuterated benzene for the lock signal (V/V = 5/1), and then solutions were pipetted into an NMR tube. The number of transients was 5000, with a relaxation delay of 10 s. The average sequence lengths in monomer units L and G, which could be calculated from the relative dyad splitting intensities of the carbonyl carbon of the lactyl-lactyl, lactyl-glycolyl, glycolyl-glycolyl, and glycolyl-lactyl signals (I GG , I GL : signal intensities of glycolyl-glycolyl and glycolyl-lactyl bonds; I LL , I LG : signal intensities of lactyl-lactyl and lactyl-glycolyl bonds at chemical shifts ~172 ppm and ~169 ppm):

Lactic/glycolic unit ratio calculation
The L/G ratio was readily and accurately calculated based on the proportion of specific peak intensities at chemical shifts 4.5-5.5 ppm as described in "Lactic/glycolic ratio calculation." Briefly, three representative 1 HNMR spectra of PLGA polymers with different L/G ratios are shown in Fig. 1, where A We analyzed three groups of L/G ratios: 50/50, 75/25, and 100/0. The lactic content in the commercial PLGA polymers was calculated by 1 HNMR spectroscopy and is displayed in Fig. 2 and Table 1. From the data set described, some notable trends were observed, although clearly a higher quantity of lots and/ or polymer molecular weights would be needed for more definitive conclusions. The first trend observed was that for L/G ratio of 50/50 ( Fig. 2A), the lactic content decreased as the molecular weight (as reported in the product certificate of analysis) increased when compared within the same and Resomer® RG 502H (14 kDa) > Resomer® RG 504H (52.9 kDa). A second trend was observed where Resomer® polymers with ester end-capping displayed a lower lactic content than their comparable polymers with acid endcapping, e.g., L/G of Resomer® RG 502 < Resomer® RG 502H. For L/G ratios of 75/25 (Fig. 2B), Resomer® polymers were found to most closely match the listed L/G ratio, e.g., Resomer® RG 756S (75.25/24.75) and Resomer® 752H (75.26/24.74). In the L/G ratio 75/25 ester endcapped group, although the lactic content did not follow a trend with its MW or inherent viscosity (i.v.), the lactic contents were lowest in these high molecular weight polymers, e.g., Resomer® RG 756S (i.v. = 0.90 dL/g, 75.25/24.75) > Expansorb® DLG 75-9E (0.95 dL/g, 73.28/26.72) > Lac-tel® DL-PLG B6007-2P (0.81 dL/g, 72.53/27.47). For polymers with an L/G ratio of 100/0 (Fig. 2C), Expansorb® DL 100-2A and Resomer® R 202H displayed exactly 100% lactic content, although surprisingly, Purasorb® PDL 02A revealed a low level of glycolic impurities (98.61/1.39). We also observed that the lactic content slightly decreased as the molecular weight of PLGA increased, which could be attributed to the synthesis of the PLGA polymer. In either the ring-opening or direct polycondensation, the PLGA product mainly undergoes three steps: chain initiation, chain growth and termination. During polymerization, glycolic monomers are more reactive and add to the growing polymer chain easier than lactic monomers [16]. As the chainlength grows, the molecular weight of the intermediate increases, leading to an increase in steric hinderance at the hydroxyl reaction site [17]. Thus, the higher the molecular weight and the longer the reaction proceeds, there may be an increase in smaller chains forming where the more reactive glycolic monomers are preferentially added compared to lactic monomers [18].

End-group analysis
There are two common types of end groups of commercial linear PLGA, terminating in carboxylic acid or aliphatic esters, with the other end of the polymer chain terminated by a single hydroxyl group. In synthesis of PLGA polymer, hydroxyl-containing compounds such as water, lactic acid, or an alcohol are used for initiators in acid end-capped polymer, while an aliphatic alcohol, such as 1-dodecanol, is used as the initiator for obtaining an ester end-capped polymer [12]. As mentioned in "End-capping analysis," in 13 CNMR spectrum, the distinct difference between acidand ester-modified end caps of PLGA is the carbon peak of the methyl group of the aliphatic chain in the ester end-cap at a chemical shift of 14 ppm. The side methyl unit of lactic monomers should appear at a chemical shift of 16 ppm, so there is a clear split between the different methyl groups [13]. As shown in Fig. 3, in the 13 CNMR spectrum of Expansorb® DLG 50-2E (B), the first methyl peak appears very high upfield (red arrow, chemical shift of 14 ppm), suggesting an ester-modified end-cap in its structure, while no signal was observed at the same point in the 13 CNMR spectrum of Expansorb® DLG 50-2A (A), suggesting an acid end-cap. All 17 commercial PLGA polymers were analyzed by 13 CNMR spectroscopy, and the data is presented in Table 1. Five polymers, including Expansorb® DLG 75-9E, Expansorb® DLG 50-2E, Resomer® RG End-capping is considered an important index influencing the behavior of PLGA hydrolysis. Compared to the PLGA polymers with an acid end-cap, ester end-capping decreases the rate of hydrolysis. This can be explained by two aspects: first, ester-modification masks the carboxyl group at the chain terminus, enhancing the hydrophobicity and decreasing water uptake of PLGA polymers [19]; second, the ester end-cap impedes the acid-mediated autocatalysis of the acid end-capped PLGA polymers. For example, Tracy et al. [20] described that the ester-capped PLGA polymer degraded 2-3 times slower in vitro and 3-4 times slower in vivo than the uncapped (acid-capped) PLGA polymer with similar molecular weights. Most research [21] shows that ester end-capped PLGA-based formulations generally display a substantially slower hydrolysis rate and slower drug release compared to PLGA with acid end-capping, but this process is obviously affected by multiple parameters.

Monomer sequence analysis
In this paper, we chose to describe blockiness by the Rc value since the result better represents the impact of G-G units on PLGA polymer hydrolysis and degradation. As mentioned in "Blockiness analysis," based on the different chemical molecular environments of the carbonyl carbons in G-G (blue) and glycolyl-lactyl (G-L, red) bonds (shown in Fig. 4), the blockiness value (Rc) can be calculated by dividing the peak intensities located at chemical shift 166-167 ppm [12]. In this cluster of two peaks, the upfield peak represents the carbonyl of a glycolic unit connecting to another glycolic unit (I G-G ), while the downfield peak (I G-L ) represents the carbonyl of a glycolic connected with a lactic unit.
As the blockiness value increases, higher numbers of glycolic monomer units are grouped together and are less connected with  13 CNMR spectra (CDCl 3 , 500 MHz) of carbonyl regions of glycolic mer units of the representative Expansorb® DLG 50-2E. The red area is intensity of the carbonyl peak of glycolic mers adjacent to lactic mers, while the blue zone represents the intensity of the carbonyl peak of a glycolic mers adjacent to another glycolic mers lactic monomers, resulting in more intrachain heterogeneity of PLGA polymers [4]. A larger number of G-G blocks would be expected to increase water uptake of PLGA (presuming the absence of crystallite formation), and in turn, increase ester bond cleavage. Thus, the blockiness could affect PLGA physicochemical properties such as degradation [22,23]. Meyer et al. [7,11] revealed that there was a difference in the hydrolysis process between random PLGA with higher heterogeneity and higher block lengths relative to more homogenous PLGA polymers with a 50/50 L/G ratio. The hydrolysis of the random and more "blocky" PLGA polymer underwent a very fast initial step due to the rapid degradation rate of the more sterically accessible G-G bonds compared to lactic-rich blocks, and then the degradation speed gradually slowed as lactic-rich blocks remained. The hydrolysis of the sequenced PLGA with low or no blockiness was steady during the whole period. These results implied that PLGA polymers with high blockiness could lead to a fast loss of molecular weight or uneven erosion in the process of degradation, which might affect the long-acting zero-order release behavior of encapsulated drugs. Therefore, as seen in Fig. 5, Resomer® RG 756S (Rc = 1.7) displayed higher blockiness than Expansorb® DLG 75-9E (Rc = 0.5) and Lactel® DL-PLG B6007-2P (Rc = 0.6), implying that the former may possess a relatively more rapid hydrolysis rate than the other two polymers.

Block length analysis
Almost all commercial PLGA polymers have a non-statistical distribution of the co-monomers, which implies that there are different block lengths of glycolic or lactic monomers. Technically, block lengths are influenced by the synthesis parameters (such as reaction type and time) which stems from the increased reactivity of glycolide compared to that of lactide [24]. The average block length of either lactic units or glycolic units, associated closely with the blockiness, could affect the physicochemical properties of PLGA polymers, such as solubility and crystallinity, leading to different degradation and hydrolysis behaviors. As the glycolic sequence length increases, the solubility of PLGA polymers has been reported to decrease in organic solvent [17]; e.g., the amorphous PLLGA (46.3 (L-lactide)/53.7) with an L G value of 4.3 was reported insoluble in chloroform at 25 °C [15]. As described in "Block length analysis" [15,25], the block lengths of both lactic and glycolic unit sequences can be calculated from the peak intensities of the dyad splitting of carbonyl carbons in the 13 CNMR spectroscopy (seen in Fig. 6). If there was a truly random distribution of lactic and glycolic units in PLGA 50/50, the theoretical values of block lengths (both L L and L G ) should be near or equal to 2 [26]. Lower values are also theoretically obtained when a "transesterification" process occurs during the reaction.
As shown in Table 1, both L L and L G of these commercial PLGA polymers were typically > 2, indicating non random distributions of the lactic and glycolic units. In the group with an L/G ratio 50/50, low molecular weight PLGA polymers (~12 kDa) including both Expansorb® DLG 50-2E (L L = 2.5, L G = 2.6) and Resomer® RG 502 (L L = 2.6, L G = 2.5) displayed lower block lengths than the other polymers evaluated. For a moderate molecular weight range (32-58 kDa), Resomer® RG 504H displayed the lowest block lengths (L L = 2.9 and L G = 2.9) compared to Expansorb® DLG 50-5A (L L = 4.0, L G = 4.0) and Purasorb® PDLG 5004A (L L = 3.4, L G = 3.2). These PLGA polymers with shorter block lengths of both lactic and glycolic units would be expected to undergo a steadier rate of hydrolysis compared to a PLGA with random heterogeneity. ). An increase in lactic block length (L L ) due to the increased proportion of lactic acid in the L/G ratio is expected, whereas the glycolic block lengths (L G ) are more informative of the PLGA sequence distribution. Wako® 7515 PLGA is produced by polycondensation (PC) in contrast with the most common ring opening polymerization (ROP) method. Here, it is interesting to point out that the Wako polymer had the lowest glycolic block length within its class. Both ROP and PC can result in long block lengths due to the increased reactivity of glycolide or glycolic acid [17], yet, it has been shown that very low block lengths can be achieved in PC by increasing the reaction time and although lower block lengths can also similarly be achieved in ROP, even a dramatically long reaction time does not necessarily result in a completely homogeneously sequenced structure [27]. Furthermore, combined with high blockiness, the long block length of glycolic units (L G ), meaning more reactive G-G bonds, would be expected to contribute to the increased hydrolytic rate in the initial step of PLGA degradation. For instance, when all other variables are considered equal or similar, which is usually difficult to achieve, Expansorb® DLG 50-2A (Rc: 1.9, L G : 4.2) would be expected to have increased initial degradation rate compared to Expan-sorb® DLG 50-5A (Rc: 0.7, L G : 4.0) and Resomer® RG 502H (Rc: 1.4, L G : 3.0).

Conclusion
Four important properties describing the chemical microstructures of 17 commercial PLGA polymers across 5 different brands and varying L/G ratio, end-capping, and molecular weights have been investigated by 1 H and 13 CNMR spectroscopy. Overall, polymers with similarly labeled properties from different manufacturers are quite comparable. It was observed that within groups of similar L/G ratio and molecular weights, different brands of PLGA polymers possessed distinct blockiness and block lengths, which could lead to diverse behaviors. Based on these measured values, it is anticipated that when other  13 CNMR spectra (hexafluoro-2-propanol/benzenze-d 6 , 500 MHz) of the carbonyl regions of PLGA polymer (representative is Expansorb® DLG 50-5A). Indicated with arrows, the areas of pink (L-G), brown (L-L), blue (G-G), and red (G-L) represent the intensi-ties of the carbonyl peak of lactic mer units adjacent to glycolic mers units, lactic adjacent to lactic, glycolic adjacent to glycolic, and glycolic adjacent to lactic, respectively parameters are in the same range, the PLGA polymers with low blockiness and shorter block lengths of both glycolic and lactic monomers may show more uniform erosion and steady degradation rates. In theory, these analytical methods could be used as a guide for polymer selection across multiple fields, especially long-acting release of drugs. Further research on the correlation between these properties and PLGA degradation process and PLGA drug release in vitro and in vivo will be reported by our lab in the future.