Effect of trehalose polymer regioisomers on protein stabilization

Marco S. Messina ab, Jeong Hoon Ko ab, Zhongyue Yang ab, M. Jane Strouse a, K. N. Houk ab and Heather D. Maynard *ab
aDepartment of Chemistry and Biochemistry, University of California, Los Angeles, 607 Charles E. Young Drive East, Los Angeles, California 90095-1569, USA. E-mail: maynard@chem.ucla.edu
bCalifornia NanoSystems Institute, University of California, Los Angeles, 570 Westwood Plaza, Los Angeles, California 90095-1569, USA

Received 25th April 2017 , Accepted 25th June 2017

First published on 3rd July 2017


There is considerable interest in the use of proteins as therapeutics and as chemical and biochemical reagents. However, many proteins are unstable and aggregate when exposed to stressors, including increased temperature, pH change, agitation, and desiccation. Polymers with side chain trehalose units were shown to be effective protein stabilizers, preventing aggregation and prolonging activity. Herein, we report the synthesis and characterization of four trehalose regioisomers containing a vinylbenzyl ether moiety at either the 2-O, 3-O, 4-O, or 6-O position. Computational analysis of these regioisomers suggested that they differ in their conformational flexibility, but all retained the native clam shell conformation of trehalose. Polymers were synthesized from the monomers separately via free radical polymerization and one polymer was prepared containing all of the regioisomers. The polymers were tested for their ability to stabilize insulin, and were found to prevent agitation-induced aggregation comparably. The results show that for insulin the effect of trehalose positional modification is minimal and suggest that the clam shell conformation itself may be more important than the polymer backbone attachment site for stabilization of proteins.


Introduction

Proteins are widely used as therapeutics in the pharmaceutical industry, feed-stock additives in the agricultural industry, and biochemical reagents in the laboratory setting. However, many proteins are prone to inactivation when exposed to outside stressors such as heat,1 pH changes,2 agitation,3 and desiccation,4 and their instability during the production, storage, and transport increases their cost.5 To prevent denaturation and thereby prolong protein activity, excipients such as sugars and polymers are often added to protein formulations.6

Trehalose, a non-reducing disaccharide formed by α,α-1,1-linked glucose units,7 is upregulated in lower-level organisms such as tardigrades during long periods of desiccation.8,9 This increase in trehalose concentration imparts stability to the organism by protecting the cell membrane and proteins.10 The mechanism of trehalose protein stabilization is under debate and there exist several different hypotheses.11–13 The three main hypotheses include water replacement,13 mechanical entrapment (vitrification),10 and water entrapment.14 In the water replacement theory, trehalose forms direct hydrogen bonds with the protein, effectively replacing water molecules and acting as the protein hydration shell. The mechanical entrapment hypothesis suggests that trehalose forms a glassy matrix around the protein, thereby reducing the mobility of the protein and allowing it to retain its tertiary structure. The water entrapment theory states that trehalose molecules trap water molecules around the protein to form a water hydration layer between the protein and trehalose. While the exact mechanism, or the combination of multiple mechanisms, responsible for the stabilization of proteins by trehalose remains to be fully determined,15 the stability that trehalose imparts on proteins remains clear. It is this feature that has enabled its use as an excipient in a range of protein therapeutic formulations such as Herceptin®, Avastin®, and Advate®.16 Trehalose has also been effective as an excipient for the stabilization of reverse transcriptase,17 as an embedding medium for preserving protein structure during electron crystallography,18 and as an additive to improve shelf-life of food/pharmaceutical/cosmetic products.16

Motivated by these features of trehalose, we developed polymeric materials based on trehalose that stabilize proteins ranging from enzymes,19–21 growth factors,22,23 hormones,24 and antibodies22,25 to various stressors including heat, lyophilization, agitation, and direct electron beam irradiation. Other groups have also used trehalose containing polymers in the prevention of amyloid beta (Aβ) aggregation26 and small interfering RNA (siRNA) and plasmid DNA (pDNA) delivery.27,28 Previously, we have explored the effect of the polymer backbone identity on the overall stabilization properties of trehalose glycopolymers by comparing polystyrene and polymethacrylate backbones as excipients to stabilize horseradish peroxidase (HRP) to heat and β-galactosidase (β-Gal) to lyophilization.20 Slight differences in stabilizing effect were observed for different polymer backbones at low equivalents of the polymer, but at higher equivalents all of the polymers stabilized the proteins, regardless of polymer backbone.

We were thus motivated to systematically investigate the effect of the point of linkage on trehalose while keeping the polymer backbone the same. To study possible differences between trehalose regioisomers on protein stabilization, we prepared styrenyl trehalose monomers with trehalose modified at the 2-O, 3-O, 4-O, or 6-O positions (Scheme 1). The resulting polymers, as well as a polymer containing all of the regioisomers, were then tested as excipients for the stabilization of the model protein insulin to mechanical agitation. The results are described herein.


image file: c7py00700k-s1.tif
Scheme 1 Synthesis of trehalose monomer regioisomers.

Experimental

Materials

Trehalose was purchased from The Healthy Essential Management Corporation (Houston, TX), dried with ethanol, and stored under vacuum. Azobisisobutyronitrile (AIBN) (98%) was purchased from Sigma-Aldrich and recrystallized from acetone before using. 4-Vinylbenzyl chloride (90%) was purchased from Sigma-Aldrich. Insulin, human recombinant (Cat. No. 91077C; Lot No. 15L255-D) was purchased from Sigma Aldrich. Sodium hydroxide (≥97%, Pellets/Certified ACS), N,N-dimethylformamide (DMF) (≥99.8%, Certified ACS), dimethyl sulfoxide (DMSO) (≥99.9%, Certified ACS), Eppendorf LoBind® microcentrifuge tubes (0.5 mL and 1.5 mL), and pyridine (≥99%, Certified ACS) were purchased from Fisher Scientific. Pyridine was dried via distillation over calcium hydride and stored over 3 Å molecular sieves. Spectra/Por® 3 dialysis membrane standard RC tubing (MWCO: 3.5 kDa) was used for dialysis of polymers. Deuterated solvents (Cambridge Isotope Laboratories) for NMR spectroscopic analyses were used as received.

Analytical techniques

NMR spectra were recorded on Bruker AV 400, 500, or DRX 500 MHz spectrometers. Chemical shifts are reported in ppm relative to the residual signal of the solvent (D2O: δ 4.79 ppm, CDCl3: δ 7.26 ppm, or (CD3)2SO: δ 2.50 ppm). 1H NMR spectra are reported as follows: chemical shift (δ ppm), multiplicity (t = triplet, d = doublet, dd = doublet of doublets, m = multiplet), coupling constant (Hz), and integration. 1H NMR spectra were acquired with a relaxation of 2 s for small molecules and 30 s for polymers with an acquisition time of 3.27 s and 30° pulse angle. Gel permeation chromatography (GPC) was conducted on a Shimadzu high performance liquid chromatography (HPLC) system with a refractive index RID-10A, one Polymer Laboratories PLgel guard column, and two Polymer Laboratories PLgel 5 μm mixed D columns. Eluent was DMF with LiBr (0.1 M) at 50 °C (flow rate: 0.80 mL min−1). Calibration was performed using near-monodisperse pMMA standards from Polymer Laboratories. HPLC purification of trehalose monomers was performed on a Shimadzu HPLC system with a refractive index and UV detector RID-10A monitoring at λ = 254 and 220 nm, and one Luna 5 μm C18(2) 100 Å LC column (250 × 21.2 mm) with 40% MeOH and 60% H2O isocratic eluent mixture at a flow rate of 20 mL min−1. The same HPLC system, equipped with an analytical Luna 5 μm C18(2) 100 Å column (250 × 460 mm), was utilized for detection of insulin with a gradient solvent system (water[thin space (1/6-em)]:[thin space (1/6-em)]acetonitrile = 30[thin space (1/6-em)]:[thin space (1/6-em)]70 to 40[thin space (1/6-em)]:[thin space (1/6-em)]60 + 0.1% trifluoroacetic acid over 15 min at 1 mL min−1). Thermogravimetric analysis (TGA) was performed on a PerkinElmer Diamond TG/DTA instrument with a ramping rate of 10 °C per minute. Infrared (IR) spectra were obtained with a PerkinElmer Spectrum One instrument equipped with a universal ATR assembly; spectra are reported in wavenumbers ([small nu, Greek, tilde]). Mass spectra were acquired on a Waters Acquity Ultra Performance Liquid Chromatography (UPLC) connected to a Waters LCT-Premier XE Time of Flight Instrument controlled by MassLynx 4.1 software. The mass spectrometer was equipped with a Multi-Mode Source operated in the electrospray mode. Trehalose samples were separated using an Acquity BEH C18 1.7 μm column (2.1 × 50 mm) and were eluted with a gradient of 5–50% or 10–45% solvent B over 6 min (solvent A: water, solvent B: acetonitrile, both with 0.2% formic acid (vol/vol)). Mass spectra were recorded in the negative ion mode in the m/z range of 70–2000 with leucine enkephalin (Sigma L9133) as the lock mass standard. Mass spectra were also collected on a Thermo Scientific Exactive Plus mass spectrometer with IonSense Direct Analysis in Real Time (DART-MS) ID-CUBE. Samples of insulin were stressed in a New Brunswick Scientific Excella E24 Incubator Shaker.

Computational methods

Conformers for regioisomer O2, O4, and O6 were searched by using Maestro 9.4 with OPLS_2015 force field in implicit water. For each regioisomer, the ensemble of conformers consists of those whose energies are within 10 kcal mol−1 from the lowest one. This ensemble typically includes ∼400 structures. The structures were then clustered to 25 representative structures for O2, 33 for O4, and 42 for O6 using the chemical informatics tool in Maestro 9.4. These structures were then optimized using B3LYP/6-31 g(d) with SMD water model in Gaussian 09.29 Frequency analysis was conducted to confirm that the structures are stationary points on the potential energy surface with no imaginary frequency. Thermal energies are calculated by using simple harmonic oscillator model. The reported energies are Gibbs free energies at 298.15 K and 1 bar.

Synthesis

6-O-(4-Vinylbenzyl ether)-α,α-trehalose (O6), 4-O-(4-vinylbenzyl ether)-α,α-trehalose (O4), 3-O-(4-vinylbenzyl ether)-α,α-trehalose (O3), 2-O-(4-vinylbenzyl ether)-α,α-trehalose (O2). NaOH (4.44 g, 1.14 × 10−1 mol) was added to DMSO (100 mL) and stirred for 5 min. Trehalose (4.86 g, 1.42 × 10−2 mol) was then added to the reaction flask. Once trehalose dissolved, 4-vinylbenzyl chloride (0.4 mL, 2.55 × 10−3 mol) was added dropwise and reaction turned yellow. The reaction was stirred for 12 hours at 25 °C and was then precipitated in a mixture of cold hexanes (1.6 L) and dichloromethane (400 mL). Precipitate was collected via filtration and dried under reduced pressure to afford a yellow-white solid. The solid was dissolved in H2O (50 mL) and neutralized with 12 N hydrochloric acid (HCl). Once neutralized, MeOH (50 mL) was added and the solution mixed. The solution was then filtered through a 0.45 μm cellulose acetate filter and purified via preparative HPLC (40% MeOH in H2O). MeOH was removed under reduced pressure and water was removed via lyophilization to afford compounds O2, O3, O4, and O6 in 11%, <1%, 39%, and 13% yield, respectively, as fluffy white powders. The combined yield for all the regioisomers was 64%.

O2: HPLC retention time (peak intensity): 10.3 minutes. 1H NMR (500 MHz in D2O, 298 K): δ = 7.47–7.45 (m, 2H), 7.35–7.33 (m, 2H), 6.77–6.71 (m, 1H), 5.84–5.80 (d, J = 17.69 Hz, 1H), 5.30–5.28 (d, J = 11.42 Hz, 1H), 5.23–5.22 (d, J = 3.68 Hz, 1H), 5.14–5.13 (d, J = 4.05 Hz, 1H), 4.69–4.61 (m, 2H), 3.91–3.51 (m, 10H), 3.45–3.39 (m, 2H); 13C NMR (125 MHz in D2O, 298 K): δ = 137.5, 136.6, 136.2, 128.9, 126.4, 114.7, 93.5, 91.4, 78.7, 73.2, 72.4, 72.2, 72.0, 72.0, 71.0, 69.8, 69.2, 60.6, 60.1; IR [small nu, Greek, tilde] (cm−1): 3294 (br), 2923, 1635, 1362, 1043, 988, 827, 803; LC-MS (±1.0) observed (predicted): [M + HCOO] 503.1762 (503.1765).

O3: HPLC retention time (peak intensity): 11.7 minutes. 1H NMR (500 MHz in D2O, 298 K): δ = 7.53–7.45 (q, 4H), 6.83–6.78 (dd, 1H), 5.88–5.85 (d, J = 17.86 Hz, 1H), 5.33–5.31 (d, J = 11.02 Hz, 1H), 5.20–5.19 (m, 2H), 4.86 (s, 2H), 3.91–3.82 (m, 6H), 3.78–3.73 (m, 3H), 3.67–3.64 (m, 1H), 3.57–3.53 (t, J = 9.63 Hz, 1H), 3.47–3.43 (t, J = 9.63 Hz, 1H); 13C NMR (125 MHz in D2O, 298 K): δ = 137.3, 136.2, 128.9, 126.2, 114.4, 93.2, 93.0, 81.3, 74.8, 72.4, 72.2, 72.1, 70.9, 70.8, 69.6, 69.3, 60.4, 60.3. IR [small nu, Greek, tilde] (cm−1): 3301 (br), 2932, 1628, 1512, 1406, 1358, 1285, 1259, 1216, 1146, 1105, 1080, 1027, 986, 943, 910, 827, 802; DART-MS observed (predicted): [M − H] 457.17040 (457.17044).

O4: HPLC retention time (peak intensity): 14.6 minutes. 1H NMR (500 MHz in D2O, 298 K): δ = 7.53–7.41 (q, 4H), 6.83–6.78 (dd, 1H), 5.89–5.85 (d, J = 17.73 Hz, 1H), 5.34–5.32 (d, J = 10.97 Hz, 1H), 5.19–5.16 (m, 2H), 4.89–4.84 (d, J = 10.81 Hz), 4.71–4.69 (d, J = 10.81 Hz, 1H), 3.99–3.95 (t, J = 9.62, 1H), 3.86–3.79 (m, 5H), 3.76–3.72 (m, 2H), 3.68–3.66 (m, 1H), 3.63–3.59 (m, 1H), 3.54–3.50 (t, J = 9.46 Hz, 1H), 3.45–3.41 (t, J = 9.62 Hz, 1H); 13C NMR (125 MHz in D2O, 298 K): δ = 137.5, 136.6, 136.2, 129.2, 126.3, 114.6, 93.2, 93.0, 77.7, 74.6, 72.6, 72.4, 72.0, 71.1, 71.1, 70.9, 69.6, 60.4, 60.2; IR [small nu, Greek, tilde] (cm−1): 3234 (br), 2930, 1629, 1360, 1107, 1043, 992, 913, 827, 805; LC-MS (± 1.0) observed (predicted): [M + HCOO] 503.1720 (503.1765).

O6: HPLC retention time (peak intensity): 19.7 minutes. 1H NMR (500 MHz in D2O, 298 K): δ = 7.52–7.38 (q, 4H), 6.82–6.76 (dd, 1H), 5.87–5.84 (d, J = 17.51 Hz, 1H), 5.32–5.30 (d, J = 11.03 Hz, 1H), 5.17–5.15 (m, 2H), 4.62–4.56 (q, 2H), 3.97–3.94 (m, 1H), 3.85–3.79 (m, 5H), 3.76–3.70 (m, 2H), 3.64–3.60 (m, 2H), 3.47–3.41 (q, 2H); 13C NMR (125 MHz in D2O, 298 K): δ = 137.3, 136.8, 136.2, 128.7, 126.3, 114.5, 93.3, 93.2, 72.6, 72.5, 72.4, 72.1, 70.9, 78.8, 70.7, 69.9, 69.6, 68.6, 60.5; IR [small nu, Greek, tilde] (cm−1): 3328 (br), 2928, 1630, 1512, 1407, 1365, 1212, 1147, 1105, 1076, 1032, 987, 942, 909, 826, 805, 718; LC-MS (± 1.0) observed (predicted): [M + HCOO] 503.1765 (503.1765).

Representative free radical polymerization (P4)

O4 (531 mg, 1.16 mmol, 33 eq.) and AIBN (5.77 mg, 35.1 μmol, 1 eq.) were dissolved in H2O (3.63 mL) and DMF (7.27 mL). The mixture was added to a dry Schlenk tube and subjected to five freeze–pump–thaw cycles. The polymerization was started by immersing the Schlenk tube in a 90 °C oil bath. The polymerization was stopped after 21 hours by cooling with liquid nitrogen and exposing the reaction to air. The polymer was purified via dialysis (MWCO = 3.5 kDa) against H2O for three days and lyophilized to produce a fluffy white solid.

P2: Mn = 1.9 kDa (Mn calculated from acetylated polymer, vide infra for details: 9.5 kDa); Mw = 2.1 kDa; Đ = 1.09; 1H NMR (500 MHz in DMSO) δ: 7.06, 6.38, 5.06, 4.90, 4.51, 4.39, 3.71, 3.51, 3.36, 3.21, 2.27–0.43; IR: [small nu, Greek, tilde] (cm−1): 3351, 2950, 1623, 1425, 1400, 1362, 1150, 1080, 991, 804. TGA weight loss onset temperature: 263 °C.

P4: Mn = 14.8 kDa (Mn calculated from acetylated polymer: 15.1 kDa); Mw = 23.2 kDa; Đ = 1.56; 1H NMR (500 MHz in DMSO) δ: 7.03, 6.40, 5.11, 4.92, 4.85, 4.78, 4.60, 4.41, 3.77, 3.65, 3.58, 3.49, 3.16, 2.11–0.66; IR: [small nu, Greek, tilde] (cm−1): 3342, 2928, 1637, 1423, 1359, 1148, 1104, 1039, 987, 846, 803, 706. TGA weight loss onset temperature: 286 °C.

P6: Mn = 1.8 kDa (Mn calculated from acetylated polymer: 9.4 kDa); Mw = 1.9 kDa; Đ = 1.05; 1H NMR (500 MHz in DMSO) δ: 7.05, 6.47, 4.90, 4.80, 4.68, 4.41, 3.89, 3.69, 3.58, 3.49, 3.17, 2.03–0.63; IR: [small nu, Greek, tilde] (cm−1): 3344, 2924, 2162, 1636, 1423, 1362, 1147, 1076, 1034, 989, 940, 847, 806, 706. TGA weight loss onset temperature: 274 °C.

PA (all isomers mixed): Mn = 4.4 kDa (Mn calculated from acetylated polymer: 14.6 kDa); Mw = 5.4 kDa; Đ = 1.21; 1H NMR (500 MHz in DMSO) δ: 7.02, 6.42, 5.07, 4.91, 4.82, 4.55, 4.41, 3.88, 3.58, 3.49, 3.16, 2.49, 2.18–0.66; IR: [small nu, Greek, tilde] (cm−1): 3348, 2922, 1430, 1362, 1041, 990, 804. TGA weight loss onset temperature: 284 °C.

Representative polymer acetylation (P4)

P4 (11.5 mg, 25.1 μmol, 1 eq.) was dissolved in dry pyridine (1.0 mL) and added to a dry and degassed round bottom flask. After five minutes of stirring, acetic anhydride (59.3 μL, 0.627 mmol, 25 eq.) was added dropwise. The solution stirred at room temperature for 48 hours. After 48 hours, the polymer was precipitated twice from cold diethyl ether. The precipitate was then collected and freeze-dried from benzene to afford product as a white powder. Deacetylated polymer molecular weights (provided above) were calculated using the following equation:
image file: c7py00700k-t1.tif

P2 -OAc: Mn = 15.7; Mw = 25.0 kDa; Đ = 1.59; 1H NMR (400 MHz in CDCl3) δ: 6.93, 6.41, 5.43, 5.28, 5.09, 5.01, 4.89, 4.57, 4.25, 4.07, 3.93, 2.09, 2.08, 2.01, 0.15, 0.14, 0.11, 0.10.

P4 -OAc: Mn = 19.6 kDa; Mw = 31.7 kDa; Đ = 1.62; 1H NMR (400 MHz in CDCl3) δ: 6.89, 6.36, 5.52, 5.44, 5.28, 5.23, 5.05, 5.02, 4.99, 4.49, 4.22, 4.05, 3.96, 3.54, 2.07, 2.05, 2.01, 1.26, 1.20.

P6 -OAc: Mn = 15.4 kDa; Mw = 21.2 kDa; Đ = 1.37; 1H NMR (400 MHz in CDCl3) δ: 6.85, 6.29, 5.47, 5.30, 5.04, 4.36, 4.25, 4.09, 4.03, 4.00, 3.50, 2.07, 2.05, 2.02, 2.02, 1.91, 1.84.

PA -OAc: Mn: 24.0 kDa; Mw = 35.0 kDa; Đ = 1.62; 1H NMR (500 MHz in CDCl3) δ: 7.55–6.03, 5.07, 4.51–4.15, 3.71, 3.53, 3.32, 2.07–0.50.

Insulin aggregation studies

A solution of insulin (1.0 mg mL−1) was prepared by dissolving insulin in Dulbecco's phosphate-buffered saline (DPBS, pH 7.4). Aliquots of the insulin solution (100 μL) were mixed with DPBS buffer (control, 100 μL) or stock solutions (100 μL) containing 1 or 10 weight equivalents of P2, P4, P6, or PA dissolved in DPBS in 1.5 mL screw-top dram vials. These samples were taped horizontally for maximum surface area and stressed at 37 °C in an incubating shaker set to 250 rpm for 3 hours. After 3 hours, the samples were removed from the shaker and placed in a 4 °C refrigerator until analytical HPLC analysis.

Results and discussion

The styrenyl trehalose monomers were synthesized using a single-step Williamson etherification. While the synthetic route does not require protecting group strategies, it does result in four regioisomeric monomers O2, O3, O4, and O6. Fortunately, the isomers exhibited significantly different retention times on the HPLC (Table 1, top), which allowed us to separate the monomers.
Table 1 HPLC trace and yields for trehalose monomer regioisomers

image file: c7py00700k-u1.tif

Monomer Isolated yield
O2 11%
O3 <1%
O4 39%
O6 13%
OA 64%


The identity of each regioisomer was assigned after extensive characterization by NMR spectroscopy (COSY, HMBC, and HSQC) (Fig. S1–S20). Although the regioisomers were expected to exhibit very similar characteristics, the coupling of the geminal benzyl protons in the 1H NMR spectra varied significantly, with O4 exhibiting strong coupling (10.8 Hz) indicative of nonequivalent geminal protons in significantly different environments and large Δδ (0.16 ppm; Fig. 1B) and O2 and O6 exhibiting similarly strong coupling (Fig. 1A and C), while O3 did not show any benzyl proton coupling (Fig. S6). This spectroscopic data gave us an indication that each monomer likely adopts a different conformation in solution. Direct NMR observation of through-space correlation in aqueous environment was not possible due to the broadening of the trehalose hydroxyl proton signals in water. Therefore, we computationally explored the differences in the aqueous conformation of the regioisomers.


image file: c7py00700k-f1.tif
Fig. 1 Benzyl region of 1H NMR for monomers O2, O4, and O6, and their corresponding lowest energy conformations in aqueous solution.

Briefly, for each isomer a conformational search was conducted using Maestro 10.4 and select conformers were optimized by density functional theory (DFT) calculation at B3LYP-D3/6-31G(d) level of theory in Gaussian 09.29 As shown by the lowest energy conformers for each isomer (Fig. 1D–F), all of the isomers retain the so-called clam shell conformation, in which the disaccharide is bent at the anomeric position, bringing the two glucose rings in close proximity that is characteristic for trehalose.30,31 All of the stable conformations (defined prior to the calculations as within 2 kcal mol−1 energy with respect to the most stable conformation) retain the clam shell conformation (Fig. S52) as opposed to the higher energy more open conformation (Fig. S53). However, O6 has a single most stable conformation within 4 kcal mol−1 (i.e., 99.9% of the population will be in this conformation at any given time according to Boltzmann distribution), and O4 has two stable conformations within 2 kcal mol−1 that only differ by 0.1 kcal mol−1 in energy. O2 has multiple stable conformations within 2 kcal mol−1. These results suggest that O6 and O4 have a relatively rigid conformation while other regioisomers are more flexible and fluctuate among the multiple low energy conformations. This result is reasonable, since O2 substitution would cause the most steric hindrance to the opposite ring due to the spatial proximity of the vinyl benzyl unit, while O6 would cause the least hindrance. Furthermore, both of the lowest-energy conformations of O4 show that one of the benzyl protons is proximal to the oxygen of the adjacent hydroxyl on C3 (2.41 and 1.92 Å for the two lowest energy conformers, (Fig. 1E)), which would explain the exceptionally large Δδ of O4 benzyl protons in the 1H NMR spectrum (Fig. 1B).

The yields for all of the regioisomers are provided in Table 1; OA denotes the combined yield of all of the monomer regioisomers. Interestingly, O4 was the most favored product. This observed regioselectivity was unexpected, as the primary hydroxyl (O6) would be anticipated as the major product in a simple SN2 reaction such as Williamson etherification. Based on literature reports of metal-trehalose ionic complexation,32,33 we hypothesized that ionic complexation of sodium with trehalose may be responsible for the reduced nucleophilicity of the primary hydroxyl. It has been reported that sugars complex with cations in the following order: Ca2+ > Mg2+ > Na+ > K+,32 and the crystal structure of Ca2+ with trehalose indicates that 2-O, 3-O, and 6-O chelate the cation.33 One would therefore expect that the use of potassium hydroxide in place of sodium hydroxide would result in a relatively looser ion pairing at 6-O and increased modification at the primary hydroxyl due to its intrinsically higher nucleophilicity, if ionic complexation were responsible for the unusual selectivity. Indeed, the yield of O6 relative to O4 was increased when potassium hydroxide was used as the base or when less sodium hydroxide was used than in the reaction (Table S1). This was further supported by the increased relative yield of O6 at higher temperature or in water, both of which would attenuate the effect of ionic complexation (Table S2). In water O6 was the major product as expected. However, the absolute yield of the monomers in water was low even in the presence of a phase transfer catalyst, which was likely due to the hydrolysis of the vinylbenzyl chloride.

Modulation of sugar hydroxyl reactivity by intramolecular hydrogen bonds34 and metal ions35 has been previously observed. Benzoylation of methyl α-D-glucopyranoside in pyridine showed that hydroxyl reactivity followed the order 6-OH > 2-OH > 3-OH > 4-OH.34 However, different reaction conditions changed the reactivity, sometimes even favoring the secondary alcohol 2-OH over the primary 6-OH when mannose was methylated in the presence of silver oxide.34 Miller et al. leveraged the calcium complexation of fructose to selectively modify the 3′-OH secondary hydroxyl of the fructose unit in a glycosyl acceptor in the presence of four primary hydroxyls in the donor and the acceptor.35 Our observation on the interesting chemical reactivity of trehalose adds to the body of work on regioselectivity of sugars.

With the monomer regioisomers assigned, we then targeted polymers made from each regioisomer monomer separately (P2, P4, and P6) and also one containing all regioisomers together (PA) (Fig. 2). The polymer containing all of the monomer regioisomers was synthesized by pooling the mixture purified from HPLC (OA). The polymer from O3 was not pursued due to the low yield of the monomer. The polymers were all synthesized using free radical polymerization with AIBN as the initiator in DMF and water mixtures at 90 °C.


image file: c7py00700k-f2.tif
Fig. 2 Polymer structures of P2, P4, P6, and PA. GPC traces with corresponding molecular weight and dispersity for acetyl protected (black) and unprotected (red) polymers. The molecular weights for the unprotected polymers are provided above the GPC traces. For P2, P6, and PA these were calculated from the GPC of the acetylated polymers.

Polymerization of O2, O6, and OA resulted in polymers (P2, P6, and PA), which ran close to the solvent elution on the GPC spectra, initially suggesting that the Mn of polymers were very small (<2 kDa; Fig. 2). We have previously observed that trehalose polymers with free hydroxyl groups can interact with GPC columns in DMF giving erroneous results.23 Thus, to better characterize the molecular weight, we acetylated each polymer (P2-OAc, P4-OAc, P6-OAc, and PA-OAc). This was accomplished by treating a small portion of the polymer in solution with excess acetic anhydride in dry pyridine and stirring at room temperature for 48 hours. The motivation was that by increasing the hydrophobicity of the polymer, the polymers from different regioisomers would be similarly solvated by the organic mobile phase (DMF) for accurate GPC analysis. Indeed, acetyl protected polymers showed larger Mn (15.4–24.8 kDa, Fig. 2), which allowed us to back-calculate the original polymer molecular weight. Since P4 did not give erroneous GPC readings as the other polymers, we used this polymer as a control to test the accuracy of estimating Mn in this manner. The Mn for P4-OAc was 24.8 kDa, which gives a calculated weight for P4 of 15.1 kDa. This result is very close to the 14.8 kDa Mn determined by GPC of P4 prior to acetylation (Fig. 2). This method provided us molecular weights between 9.4–14.8 kDa for all the polymers. We also tested to see if the mixture of monomers could be acetylated first and then polymerized. Indeed, the polymerization proceeded smoothly to yield PA-OAc that was subsequently deprotected as PA (see ESI for details).

We then utilized the polymers to prevent aggregation of a protein during heating to body temperature and agitation, since this is one way therapeutic proteins are degraded. Insulin was employed as the model protein, since it is an important and widely used therapeutic protein for the treatment of diabetes. Insulin solutions are prone to aggregation when agitated, which makes transportation and storage difficult.3 Inactivation of insulin, even in small amounts, poses a risk to patients due to improper insulin dosage.36

Samples were prepared of insulin and polymer at 1 and 10 weight equivalents (wt equiv.) in DPBS in 1.5 mL screw-top dram vials. The protein samples were stressed at 37 °C with 250 rpm agitation for 3 hours. Using this method, large insoluble insulin aggregates were visually observed (Fig. S50). In order to quantify the amount of intact (non-aggregated) insulin, we utilized HPLC. HPLC has been frequently employed to separate degradation products from the protein thereby enabling accurate quantification of intact insulin.37 After stress, samples were filtered through 0.2 μm syringe filters to ensure removal of insulin aggregates and analyzed via HPLC.

None of the polymers prevented aggregation at 1 wt equiv. (Fig. S51). However, all of the polymers prevented aggregation of insulin similarly (97–100%) at 10 wt equiv., whereas samples of insulin stressed without added polymer or with 10 wt equiv. of trehalose aggregated completely (Fig. 3) meaning that zero percent intact insulin was observed by HPLC. This agrees with previous observations that trehalose in polymeric form is a better stabilizer than trehalose.19,20,26 Since there were no statistical differences in stabilization between polymers at 10 wt equiv., we conclude that the trehalose monomer regioisomers can be combined to achieve higher monomer yield, and all polymers can be utilized interchangeably, at least with the protein insulin.


image file: c7py00700k-f3.tif
Fig. 3 Stabilization of insulin using 10 wt equiv. of polymer. Samples were heated to 37 °C with 250 rpm agitation for 3 hours. Intact insulin quantified by HPLC. There is no statistical difference between polymers (n = 3). Note that DPBS and trehalose have zero intact insulin %.

The computational studies have shown that while there seem to be differences in conformational flexibility between the monomer regioisomers, all of the stable conformations still possess the clam shell conformation of trehalose (Fig. S52), and it is mostly the vinyl benzyl substituent that moves in the conformations for each isomer. Studies have pointed to the axial α,α-(1→1) linkage that results in the clam shell conformation as being important for the protective ability of trehalose.30,31 Indeed, we have observed that trehalose polymers have superior protein stabilizing ability over polymers from other sugars such as lactose that have more open conformations.23 More thorough investigation is needed in the future to conclusively decouple the effects of conformational rigidity and the clam shell conformation on protein stabilization; in other words, more work will need to be done to determine if it is the clam shell conformation and the spatial arrangement of the hydroxyl groups itself or the molecular rigidity that results from the clam shell that is responsible for the stabilization. Nonetheless, we observe that the site of attachment of the trehalose to the polymer backbone does not have significant influence on the stabilizing ability. It should also be noted that the trehalose polymer stabilizes better than trehalose, likely due to the cluster glycoside effect from increased local concentration23,38 and/or the nonionic surfactant character of the hydrophilic sugar side chain attached to the hydrophobic backbone.20,23 Together, our findings offer an interesting view on the synthesis of trehalose monomers and provide us with data suggesting that monomer regioisomers can be pooled to increase trehalose polymer yields without reducing protein stabilization ability.

Conclusions

In conclusion, we synthesized four trehalose regioisomers containing an ether-linked styrene moiety positioned at the 2-O, 3-O, 4-O, or 6-O position of trehalose. The substitution position of each monomer was rigorously identified via NMR spectroscopy. NMR data suggested that each regioisomer adopted a distinct conformation in solution and computational methods were employed to explore this. Calculations gave insight into the relative rigidity of the trehalose regioisomers in solution, with monomers O6 and O4 being the least flexible with only one or two stable conformations, and monomer O2 showing multiple stable conformations suggesting that it is conformationally flexible. Despite the differences in conformational flexibility, all monomer regioisomers retained the native clam shell conformation of trehalose. We then probed the stabilization capability of each trehalose regioisomer in polymeric form. Polymers containing each monomer separately and one containing all monomer regioisomers together were synthesized via free radical polymerization. The stabilization capability of the polymers as excipients against mechanical agitation with moderate heating was then tested using insulin as a model protein. There was no substantial difference in the stabilization capability between each polymer; the different polymers prevented protein aggregation (>97%) while there was no intact insulin with trehalose itself or free protein. We conclude that different regioisomers may be combined to achieve higher yields of the polymer material while being able to effectively stabilize proteins, at least insulin, to mechanical stress.

Acknowledgements

H. D. M. thanks the NSF (CHE-1507735) for funding. M. S. M. thanks the NSF Bridge-to-Doctorate (HRD-1400789) and the Predoctoral (GRFP) (DGE-0707424) Fellowships and UCLA for the Christopher S. Foote Fellowship. The AV 500 NMR data was obtained on equipment supported by the NSF (CHE-1048804). The authors would like to thank Dr En-Wei Lin for the synthesis of P4-OAc, and Dr Peter Dornan and Professor Mike Jung for the helpful discussion on the origin of regioselectivity in the monomer synthesis. K. N. H. thanks NSF (CHE-1361104) for financial support. Computational resources were provided by the UCLA Institute for Digital Research and Education (IDRE) and the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by the NSF (OCI-1053575).

Notes and references

  1. M. Oobatake and T. Ooi, Prog. Biophys. Mol. Biol., 1993, 59, 237–284 CrossRef CAS PubMed.
  2. E. Y. Chi, S. Krishnan, T. W. Randolph and J. F. Carpenter, Pharm. Res., 2003, 20, 1325–1336 CrossRef CAS.
  3. V. Sluzky, J. A. Tamada, A. M. Klibanov and R. Langer, Proc. Natl. Acad. Sci. U. S. A., 1991, 88, 9377–9381 CrossRef CAS.
  4. W. Wang, Int. J. Pharm., 2000, 203, 1–60 CrossRef CAS PubMed.
  5. J. R. Kintzing, M. V. F. Interrante and J. R. Cochran, Trends Pharmacol. Sci., 2016, 37, 993–1008 CrossRef CAS PubMed.
  6. T. J. Kamerzell, R. Esfandiary, S. B. Joshi, C. R. Middaugh and D. B. Volkin, Adv. Drug Delivery Rev., 2011, 63, 1118–1159 CrossRef CAS PubMed.
  7. N. Teramoto, N. D. Sachinvala and M. Shibata, Molecules, 2008, 13, 1773–1816 CrossRef CAS PubMed.
  8. P. Westh and H. Ramløv, J. Exp. Zool., 1991, 258, 303–311 CrossRef CAS.
  9. L. M. Crowe, D. S. Reid and J. H. Crowe, Biophys. J., 1996, 71, 2087–2093 CrossRef CAS PubMed.
  10. J. H. Crowe, J. F. Carpenter and L. M. Crowe, Annu. Rev. Physiol., 1998, 60, 73–103 CrossRef CAS PubMed.
  11. N. Katyal and S. Deep, Phys. Chem. Chem. Phys., 2014, 16, 26746–26761 RSC.
  12. S. Paul and S. Paul, J. Phys. Chem. B, 2015, 119, 1598–1610 CrossRef CAS PubMed.
  13. G. Moiset, C. A. López, R. Bartelds, L. Syga, E. Rijpkema, A. Cukkemane, M. Baldus, B. Poolman and S. J. Marrink, J. Am. Chem. Soc., 2014, 136, 16167–16175 CrossRef CAS PubMed.
  14. R. D. Lins, C. S. Pereira and P. H. Hünenberger, Proteins: Struct., Funct., Bioinf., 2004, 55, 177–186 CrossRef CAS PubMed.
  15. N. K. Jain and I. Roy, Protein Sci., 2009, 18, 24–36 CAS.
  16. S. Ohtake and Y. J. Wang, J. Pharm. Sci., 2011, 100, 2020–2053 CrossRef CAS PubMed.
  17. P. Carninci, Y. Nishiyama, A. Westover, M. Itoh, S. Nagaoka, N. Sasaki, Y. Okazaki, M. Muramatsu and Y. Hayashizaki, Proc. Natl. Acad. Sci. U. S. A., 1998, 95, 520–524 CrossRef CAS.
  18. T. Hirai, K. Murata, K. Mitsuoka, Y. Kimura and Y. Fujiyoshi, J. Electron Microsc., 1999, 48, 653–658 CrossRef CAS PubMed.
  19. R. J. Mancini, J. Lee and H. D. Maynard, J. Am. Chem. Soc., 2012, 134, 8474–8479 CrossRef CAS PubMed.
  20. J. Lee, E.-W. Lin, U. Y. Lau, J. L. Hedrick, E. Bat and H. D. Maynard, Biomacromolecules, 2013, 14, 2561–2569 CrossRef CAS PubMed.
  21. J. Lee, J. H. Ko, E.-W. Lin, P. Wallace, F. Ruch and H. D. Maynard, Polym. Chem., 2015, 6, 3443–3448 RSC.
  22. E. Bat, J. Lee, U. Y. Lau and H. D. Maynard, Nat. Commun., 2015, 6, 6654 CrossRef CAS PubMed.
  23. E. M. Pelegri-O'Day, S. J. Paluck and H. D. Maynard, J. Am. Chem. Soc., 2017, 139, 1145–1154 CrossRef PubMed.
  24. Y. Liu, J. Lee, K. M. Mansfield, J. H. Ko, S. Sallam, C. Wesdemiotis and H. D. Maynard, Bioconjate Chem., 2017, 28, 836–845 CrossRef CAS PubMed.
  25. U. Y. Lau, S. S. Saxer, J. Lee, E. Bat and H. D. Maynard, ACS Nano, 2015, 10, 723–729 CrossRef PubMed.
  26. M. Wada, Y. Miyazawa and Y. Miura, Polym. Chem., 2011, 2, 1822–1829 RSC.
  27. A. Sizovs, L. Xue, Z. P. Tolstyka, N. P. Ingle, Y. Wu, M. Cortez and T. M. Reineke, J. Am. Chem. Soc., 2013, 135, 15417–15424 CrossRef CAS PubMed.
  28. Z. P. Tolstyka, H. Phillips, M. Cortez, Y. Wu, N. Ingle, J. B. Bell, P. B. Hackett and T. M. Reineke, ACS Biomater. Sci. Eng., 2015, 2, 43–55 CrossRef PubMed.
  29. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. J. Bearpark, J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, Gaussian 09, Gaussian, Inc., Wallingford, CT, USA, 2009 Search PubMed.
  30. F. Albertorio, V. A. Chapa, X. Chen, A. J. Diaz and P. S. Cremer, J. Am. Chem. Soc., 2007, 129, 10567–10574 CrossRef CAS PubMed.
  31. K. Sakakura, A. Okabe, K. Oku and M. Sakurai, J. Phys. Chem. B, 2011, 115, 9823–9830 CrossRef CAS PubMed.
  32. K. Oku, M. Kurose, M. Kubota, S. Fukuda, M. Kurimoto, Y. Tujisaka and M. Sakurai, Biosci., Biotechnol., Biochem., 2005, 69, 7–12 CrossRef CAS PubMed.
  33. T. Fujimoto, K. Oku, M. Tashiro and T. Machinami, J. Carbohydr. Chem., 2006, 25, 521–532 CrossRef CAS.
  34. J. Lawandi, S. Rocheleau and N. Moitessier, Tetrahedron, 2016, 72, 6283–6319 CrossRef CAS.
  35. G. Pelletier, A. Zwicker, C. L. Allen, A. Schepartz and S. J. Miller, J. Am. Chem. Soc., 2016, 138, 3175–3182 CrossRef CAS PubMed.
  36. R. Pryce, Br. Med. J., 2009, 338, a2218 CrossRef CAS PubMed.
  37. A. Oliva, J. Fariña and M. Llabrés, Int. J. Pharm., 1996, 143, 163–170 CrossRef CAS.
  38. J. J. Lundquist and E. J. Toone, Chem. Rev., 2002, 102, 555–578 CrossRef CAS PubMed.

Footnotes

Electronic supplementary information (ESI) available: Synthesis of polymers, monomer and polymer characterizations (NMR, IR, GPC, TGA), computational methods, and coordinates and energies of computed structures. See DOI: 10.1039/c7py00700k
These authors contributed equally.

This journal is © The Royal Society of Chemistry 2017