Quantitative Detection of PEGylated ... - ACS Publications

Feb 29, 2016 - R. Scott Prosser,*,†,‡ and Peter M. Macdonald*,†. †. Department of Chemical and Physical Sciences, University of Toronto Missis...
1 downloads 0 Views 1MB Size
Article pubs.acs.org/ac

Quantitative Detection of PEGylated Biomacromolecules in Biological Fluids by NMR Rohan D. A. Alvares,† Advait Hasabnis,† R. Scott Prosser,*,†,‡ and Peter M. Macdonald*,† †

Department of Chemical and Physical Sciences, University of Toronto Mississauga, Mississauga, Ontario Canada L5L 1C6 Department of Biochemistry, University of Toronto, Toronto, Ontario Canada M5S 1A8



S Supporting Information *

ABSTRACT: The accumulation, biodistribution, and clearance profiles of therapeutic agents are key factors relevant to their efficacy. Determining these properties constitutes an ongoing experimental challenge. Many such therapeutics, including small molecules, peptides, proteins, tissue scaffolds, and drug delivery vehicles, are conjugated to poly(ethylene glycol) (PEG) as this improves their bioavailability and in vivo stability. We demonstrate here that 1H NMR spectroscopy can be used to quantify PEGylated species in complex biological fluids directly, rapidly, and with minimal sample preparation. PEG bears a large number of spectroscopically equivalent protons exhibiting a narrow NMR line width while resonating at a 1H NMR frequency distinct from most other biochemical signals. We demonstrate that PEG provides a robust signal allowing detection of concentrations as low as 10 μg/mL in blood. This PEG detection limit is lowered by another order of magnitude when background proton signals are minimized using 13Cenriched PEG in combination with a double quantum filter to remove 1H signals from non-13C-labeled species. Quantitative detection of PEG via these methods is shown in pig blood and goat serum as examples of complex biological fluids. More practically, we quantify the blood clearance of 13C-PEG and PEGylated-BSA (bovine serum albumin) following their intravenous injection in live rats. Given the relative insensitivity of line width to PEG size, we anticipate that the biodistribution and clearance profiles of virtually any PEGylated biomacromolecule from biological fluid samples can be routinely measured by 1H NMR without any filtering or treatment steps.

W

Currently, a variety of methods are available to detect PEGylated moieties in biological samples including colorimetric, chromatographic, and radiolabeled techniques, as well as bioactivity and enzyme-linked immunosorbent (ELISA) methods.5 While relatively low concentrations of PEG are detectable with most such methods, they often require complex sample preparation. NMR, although a robust quantitative method, has been little used to quantify therapeutics or drug delivery systems since these species are normally present at concentrations below the practical NMR detection limit or their signals are obscured by background signals, a problem that is most acute in vivo. In this study, we demonstrate the use of 1H NMR to detect and quantify PEGylated species in biological fluids. Although conventionally such species are not visible using NMR, the chemico-physical properties of PEG produce a single, narrow, intense 1H NMR resonance at a chemical shift that is readily resolved from background 1H NMR signals, even when conjugated to a biomacromolecule, such as a protein.7

hile small molecules still dominate the pharmaceutical market, biologics, referring to any medicine produced from biological sources, constitute nearly 20% of the market and are the fastest growing segment.1,2 In many cases, biologics and, in particular, therapeutic proteins are chemically modified by the addition of poly(ethylene glycol) (PEG) to improve their pharmacological properties. PEG is nontoxic and reduces antigenicity, immunogenicity, proteolytic degradation, and renal clearance while improving stability, shelf life, and the solubility of entities to which it is attached.3,4 As such, it has been used to enhance the efficacy of small molecule, peptide, and protein therapeutics, while also improving the performance of polymeric nanoparticles, liposomes, and micellar drug delivery systems.5 At least 10 PEGylated protein conjugates have been approved for clinical use, while many more are in clinical trials.6 Pharmacokinetic studies, essential for development of therapeutics and clinical approvals, require quantitative detection of PEGylated conjugates in biological fluids. We demonstrate here that 1H NMR spectroscopy, alone or in combination with 13C-labeling of PEG, can be used to quantify PEGylated species in complex biological fluids directly, rapidly, sensitively, and with minimal sample preparation. © XXXX American Chemical Society

Received: December 1, 2015 Accepted: February 29, 2016

A

DOI: 10.1021/acs.analchem.5b04565 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Detector and a Waters styragel 5 μm HR column. Samples were prepared in THF, and the system calibrated with PMMA standards. PEG-BSA Conjugation. PEG was randomly conjugated to available BSA amines (Nα amino group and lysine residues) via reductive amination of monofunctionalized PEG aldehydes.11 PEG Aldehyde Synthesis. Aldehyde functionalized PEG was produced in a two-step process, wherein a protected aldehyde (acetal) was first attached to the PEG alcohol end groups, followed by a deprotection step to yield the aldehyde.12 PEG5k (3.008 g) was first dried by azeotropic distillation using ∼40 mL of toluene at 60 °C. Potassium tert-butoxide (0.820 g), 4chlorobutyraldehyde diethyl acetal (0.8 mL) and anhydrous toluene (26 mL) were added under nitrogen gas, and the mixture stirred for 12 h at 105 °C. After removing the solvent by distillation, the residue was dissolved in a minimum volume of DCM (7 mL) and added dropwise to cold diethyl ether (150 mL) to precipitate the resulting PEG butyraldehyde diethyl acetal (PEG5k-BDA), which was isolated by centrifugation (4000 rpm, 5 min). The DCM/diethyl ether washing step was repeated twice more. The final precipitate was dried under vacuum, yielding 3.367 g of product (containing both PEG5kBDA and potassium tert-butoxide). Acid-catalyzed hydrolysis of the PEG5k-BDA acetal was used to produce PEG butyraldehyde (PEG5k-BA). The above PEG5kBDA product (3.337 g) was dissolved in Milli-Q water (60 mL), the pH was adjusted to 3.0 using ∼4 mL of 5% phosphoric acid and the solution was stirred for 14 h at room temperature. The hydrolysis was quenched by adding NaCl (3.052 g) and adjusting the pH to 6.8 using 6 mL of 1 M NaOH. PEG5k-BA was extracted into chloroform (3 × 60 mL), the organic layers were combined, and the solvent was removed under reduced pressure. The residue was dissolved in toluene, dried by azeotropic distillation, and further purified using the above DCM/cold diethyl ether washing step. The final precipitate was dried under reduced pressure giving 2.476 g of PEG5k-BA. The structure was confirmed by NMR. PEG Conjugation to BSA. BSA (1.209 g; 18.3 μmol), PEG5kBA (0.8514 g; 181 μmol), and sodium cyanoborohydride (48 mg; 763 μmol) were dissolved in 11.4 mL of phosphate buffered saline (PBS = 50 mM phosphate, 150 mM NaCl, pH 7.33) and stirred at room temperature. To push the reaction to completion, a further 48 mg of sodium cyanoborohydride was added at each of 4, 16, and 24 h after initiation of the reaction. After stirring for a further 24 h, the reaction mixture was diluted 2−4-fold, and the putative PEG5k-BSA conjugate purified using size exclusion chromatography. Size exclusion chromatography was conducted using a Superdex 200 column (GE Healthcare). PEG5k-BSA (5 mL containing either 25 or 50 mg/mL of BSA by weight) was injected onto the column and eluted with PBS, collecting 3 mL fractions. Protein elution was monitored at 280 nm, while the PEG content of each fraction was monitored with 1H NMR spectroscopy using dioxane (0.32 mg/mL) as an integration standard. The degree of conjugation of PEG5k-BSA (an average of 6.2 PEGs tethered to each BSA molecule), and the concentration of the conjugate for later injection into rats, could be determined from the combination of UV−visible absorbance (BSA molar extinction coefficient of 42961 M−1 cm−1 at 280 nm13) and 1H NMR for BSA and PEG5k, respectively. The desired fractions were pooled and concentrated (Amicon Ultra-15 Centrifugal Filter Units, 3000 MWCO, EMD Millipore) via centrifugation. The product so

Moreover, PEG’s 1H NMR signal intensity is directly proportional to the length of the PEG chain, even up to large PEG molecular weights, that is, 1 MDa, meaning that the larger the PEG tag, the easier it is to detect the PEGylated species. As we demonstrate here, this alone allows 1H NMR detection and quantitation of PEGylated molecules at PEG concentrations in the low μg/mL range in blood or serum. However, we show further that, by enriching PEG with 13C and filtering the detected 1H signal through the 13C-labeled nuclei, the 1H NMR background signal is reduced, such that the limit of detection is decreased by a further order of magnitude. Moreover, we show that these NMR methods are readily used to evaluate blood clearance of PEG, 13C-PEG and a model PEGylated protein following their intravenous injection in rats.



MATERIALS AND METHODS Materials. Sodium cyanoborohydride, potassium tertbutoxide, the anhydrous solvents DMSO, toluene, tert-butanol, and D2O (99.9%) were purchased from Sigma-Aldrich (Oakville, ON, Canada), as were PEG chains of average weights 5 kDa (PEG5k) and 20 kDa (PEG20k). 100 kDa PEG (PEG100k) was obtained from Alfa Aesar (Heysham, United Kingdom). 13C-Ethylene oxide was purchased from SigmaAldrich (ISOTEC, Old Bridge, NJ, U.S.A.). HCl, chloroform, dichloromethane (DCM), and diethyl ether were obtained from Caledon Laboratories Ltd. (Georgetown, ON, Canada). NaCl, NaOH, and toluene were bought from EMD Chemicals Inc. (Gibbstown, NJ, U.S.A.), 4-chlorobutyraldehyde diethyl acetal from AK Scientific Inc. (Mountain View, CA, U.S.A.), phosphoric acid from Mallinckrodt Chemical Works (St. Louis, MO, U.S.A.), bovine serum albumin (BSA) from Bioshop Canada Inc. (Burlington, ON, Canada), and goat serum from Gibco Life Technologies (Burlington, ON, Canada). Pig blood was obtained at Sunnybrook Hospital (Toronto, ON, Canada) and collected directly into heparinized tubes. 13 C-PEG Synthesis. Uniformly-13C-isotopically labeled PEG (13C-PEG) was synthesized by anionic ring opening polymerization of 13C-ethylene oxide (EO).8,9 Briefly, EO (2.62 g) was dried over sodium metal and transferred to a polymerization flask, to which anhydrous DMSO (8.28 g) and potassium tert-butoxide (5.4 mg) were then added. The solution was stirred at 50 °C for 96 h, after which the reaction was quenched with 0.1 M HCl. 13C-PEG was isolated by dropwise addition of the reaction mixture into 200 mL of cold diethyl ether. The precipitated 13C-PEG, which formed immediately upon contact with the diethyl ether, was separated by centrifugation at 4000 rpm for 10 min. The supernatant was decanted and the pellet was washed three times with cold diethyl ether. The reaction proceeded essentially to completion, with 80% recovery (1.82 g) of 13C-PEG. The molecular weight of the 13C-PEG was determined by 1H NMR end group analysis, where the relative ratio of alcohol and tert-butyl end groups to PEG methylenes indicated a mean size of ∼27.6 kDa (herein after referred to as 13C-PEG28k). Although our goal was to produce monofunctionalized PEG, with flanking t-butyl and alcohol end groups, the 1H NMR spectrum indicated a considerably higher ratio (∼4:1) of alcohol to t-butyl groups. This was likely the result of trace quantities of water producing premature termination of polymerization,10 one consequence of which is a higher dispersion of polymer sizes, as indicated by a dispersity (Đ) of 1.37. The latter value was measured using GPC with a Waters 515 HPLC pump equipped with a Viscotek VE 3580 RI B

DOI: 10.1021/acs.analchem.5b04565 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

was 12 kHz, the 90° pulse width 8 μs, the acquisition time 2 s and the recovery time 59 s (sufficient to achieve quantitative integrals). A water presaturation pulse of 2 s was applied prior to the excitation, while 4−8 transients were accumulated. 1 H NMR 13C-PEG Calibration Curve. 13C-PEG28k was added to goat serum at various concentrations up to 500 μg/mL. Again, 50 μL of each of these prepared concentrations was diluted to 400 μL with D2O containing 13C-ethylene glycol at a concentration of either 150 μM (for 13C-PEG28k concentrations of 50 μg/mL and above) or 3.0 μM (for lower 13C-PEG28k concentrations). 1H NMR spectra were acquired using both a direct pulse-acquire and a standard heteronuclear multiple quantum coherence (HMQC) sequence. Acquisition parameters were similar to those in the 1H NMR experiment above, except the acquisition time was 1.5 s and the recovery time 16.5 s. Additionally, for the HMQC, the 13C 90° pulse width was 17.5 μs, and the preparation (τ) and mixing (δ) delays were 3.47 ms (=1/2J CH ) and 0 ms, respectively. A water presaturation pulse of 2 s was applied prior to the excitation, while broadband 13C decoupling was employed during acquisition. The number of transients accumulated varied between 16 and 1920. 1 H NMR PEG5k-BSA Calibration Curve. PEG5k-BSA was added to goat serum at various concentrations up to ∼180 μM BSA, as determined via UV−vis absorbance at 280 nm using the literature value extinction coefficient.13 A 50 μL aliquot of each of these prepared concentrations was diluted to 400 μL with D2O containing DMF at a concentration of 6.74 mM. 1H NMR spectra were recorded using a direct pulse-acquire sequence. Typically, the spectral width was 12 kHz, the 90° pulse width 8.8 μs, the acquisition time 2 s, and the recovery time 60 s (sufficient to achieve quantitative integrals). A water presaturation pulse of 2 s was applied prior to the excitation, while 4−8 transients were collected. Diffusion NMR. Solutions for NMR diffusion analysis were prepared as follows: PEG5k, PEG20k, PEG100k and PEG5k-BSA were all prepared at a concentration of 1 mg/mL in 98/2, D2O/ PBS, while 13C-PEG28k was at a concentration of 6.3 mg/mL D2O, with BSA at 25 mg/mL D2O. The diffusion coefficients were measured using the pulsed field gradient stimulated echo (PFG STE) 1H NMR sequence15,16 on a Varian Inova 600 MHz spectrometer equipped with a HCN cryogenic probe (Agilent Inc., Santa Clara, CA). Typically, the spectral width was 8.5 kHz, the 1H 90° pulse length 8.5 μs, the spin echo delay 8 ms, the gradient pulse duration 6 ms, the mixing delay 150 ms, and the number of transients and steady state scans both equal to 16. The gradient amplitude was arrayed over 15 values, such that a 90% decrease in signal intensity was usually achieved. Gradients were calibrated using a 99.9% D2O sample, with a reported diffusion coefficient of 1.935 × 10−9 m2 s−1,17 which yielded a maximum gradient amplitude of 382 G/cm. For STE PFG 1H NMR diffusion measurements, the resonance intensity, I, of the species of interest decreases as a function of increasing magnitude of the applied gradient pulses according to eq 1:

concentrated contained mainly PEG5k-BSA, with some higher order BSA molecules (e.g., dimers), but only insignificant amounts of unconjugated PEG5k. Relative to BSA (50 mg/mL) and PEG5k (50 mg/mL), the PEG5k-BSA fraction eluted first from the size exclusion column (Figure S1), as expected for a PEG-BSA conjugate. SDS-PAGE and diffusion NMR (see below) further confirmed PEG5k-BSA conjugation. PEG5k-BSA and BSA were run on a 10% SDS-PAGE gel, with an identical BSA content (10 μg) in each well. Precision Plus Protein Dual Xtra Standards (Bio-Rad) were used as the molecular weight protein ladder. Images were collected using a Molecular Imager Gel Doc XR+ System (Bio-Rad). Disappearance of the BSA band at 66 kDa, with the concomitant appearance of a higher molecular weight band, indicated reaction success (Figure S1). PEG NMR Calibration Curves. Known quantities of PEG5k were added to pig blood, and 13C-PEG28k and PEG5k-BSA to goat serum, to establish the validity of the NMR analysis in complex biological fluids. Choice of Integration Standard. Quantitative detection by NMR requires a judicious choice of an NMR integration standard. Both external and internal calibration methods may be employed, as reviewed recently.14 Internal calibration, involving the addition of a known amount of a given calibrant to the sample, remains the most widely used method and is conceptually straightforward. An ideal standard should be noninteracting, nonreactive, water-soluble and possess a chemical shift easily distinguished from other resonances present. For non- 13 C-labeled PEG, dimethylformamide (DMF) was found to satisfy these requirements, in particular exhibiting a chemical shift (∼7.90 ppm) in a relatively deserted region of the 1H NMR spectrum, distinct from that of water (∼4.6 ppm) or PEG (∼3.6 ppm). Dioxane (∼3.5 ppm) was also used in certain instances. For 13C-PEG, the proton chemical shift of the standard is less critical, so long as it is not too close to that of PEG or water, since background proton resonances will be largely filtered out. However, the 13C chemical shift of the standard must be close enough to that of PEG so that both can be uniformly excited by 13C rf pulses, and uniformly 13C-decoupled during acquisition of the 1H NMR signals. Moreover, the standard would have, ideally, a one-bond carbon−proton scalar coupling constant (JCH) similar to that of PEG, to minimize the formation of undesirable magnetization during the filter that might otherwise distort the appearance of the standard’s resonance. Deviation from these conditions would produce loss of signal intensity for the internal standard. Consequently, 13C-carbonyl-labeled DMF, although commercially available, is less than ideal as a 13C-filtered integration standard since both its 13C chemical shift (167.6 ppm) and JCH (196 Hz) are considerably different from those of 13C-PEG (72.2 ppm, 144 Hz). While other candidates are possible, we found 13C-ethylene glycol to be ideal, given its 1H and 13C chemical shifts of 3.54 and 65.1 ppm, respectively, and JCH of 140 Hz. 1 H NMR PEG Calibration Curve. PEG5k was added to heparinized pig blood at various concentrations up to approximately 4000 μg/mL. To mimic blood sampling studies with rats, 50 μL of each of these prepared concentrations was diluted to 400 μL with D2O containing 0.50 mg/mL DMF as the internal integration standard. 1H NMR spectra were recorded by direct pulse-acquire detection on a Varian VNMRS 600 MHz spectrometer using a H(F)CN probe (Agilent Inc., Santa Clara, CA). Typically, the spectral width

⎡ ⎛ δ ⎞⎤ I = I0* exp⎢ −(γδg )2 D⎜Δ − ⎟⎥ ⎝ ⎣ 3 ⎠⎦

(1)

where D (m2 s−1) is the diffusion coefficient, Δ (s) is the experimental diffusion time, γ (rad T−1 s−1) is the relevant magnetogyric ratio, g (T m−1) is the gradient pulse amplitude, and δ (s) is its duration. The resonance intensity in the absence C

DOI: 10.1021/acs.analchem.5b04565 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Detection of PEG in biological fluids using NMR. Left-hand panel (a−c) shows 1H NMR spectra from a 50 μL aliquot of pig blood containing the indicated concentration of PEG5k diluted to 400 μL with D2O containing 0.50 mg/mL DMF. Right-hand panel (d−f) shows 1H NMR spectra from a 50 μL aliquot of goat serum containing the indicated concentration of 13C-PEG28k diluted to 400 μL with D2O containing 3.0 μM 13 C-ethylene glycol. The spectral insets in the left-hand panel show the background signal from blood (red) relative to that of PEG5k and the DMF integration standard. The spectral inset in the right-hand panel shows the respective proton peaks for 13C-PEG28k (3.58 ppm) and 13C-ethylene glycol (3.54 ppm). In the series on the right, the 1H NMR spectra of serum containing 13C-PEG28k compares the results obtained (e) with direct pulse-acquire detection, and (f) with HMQC filtration to remove background signals from non-13C-labeled species.

of field gradients, I0*, includes the effects of longitudinal and transverse relaxation. A species exhibiting normal Gaussian

Rat Clearance Studies. All animal experiments were approved by the institutional research ethics board and performed in accordance with institutional animal care protocols. Male Sprague−Dawley rats (Charles River, Saint Constant, QC, Canada), each weighing approximately 250 g, were intravenously injected in the tail vein with one of four different PEG-containing solutions, at the indicated concentration, where the injected volume corresponded to 3.1 mL/kg body weight: PEG20k (50 mg/mL, PBS), PEG100k (50 mg/mL, PBS), 13 C-PEG28k (5.7 mg/mL, PBS) and PEG5k-BSA (75 mg/mL of BSA ≈ 33 mg PEG/mL PBS). Subsequently, 50 μL of blood was withdrawn at the site of injection from each rat at various time points (typically 3, 7, 10, 15, 35, 55, 125, 245, 365, and 1440 min) postinjection. The rats were restrained for the duration of the injection and while blood sampling. NMR samples were prepared by simply diluting the withdrawn 50 μL of blood with 350 μL of D2O containing the integration standard and a modest amount of heparin (∼5 units/mL) to

( ) versus

diffusion will have a linear semilogarithmic plot of ln

(

k = − (γδg )2 Δ − 15

δ 3

I I0

) with a slope proportional to its diffusion

coefficient. Longitudinal and Transverse Relaxation Time Measurements. The PEG 1H NMR spin−lattice (T1) and spin− spin (T2) relaxation times were measured at room temperature for PEG5k, PEG20k, PEG100k, 13C-PEG28k, and PEG5k-BSA, all prepared at a concentration of 5 mg/mL in D2O, on a Varian VNMRS 600 MHz spectrometer using a H(F)CN probe (Agilent Inc., Santa Clara, CA) using the inversion−recovery and spin−echo techniques, respectively. Typical parameters were a 1H pulse width of 8.35 μs, a spectral width of 12 k, an acquisition time of 2 s, and a total relaxation time of 77 s. The T1 or T2 delay was arrayed over 23 values, sufficient to capture decays for all species present. The spectra for 13C-PEG28k were acquired under broadband 13C decoupling. D

DOI: 10.1021/acs.analchem.5b04565 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. Heteronuclear multiple quantum coherence (HMQC) filter. After excitation of the protons, the magnetization is transferred to adjacent 13 C nuclei using the first carbon pulse placed at an appropriate durations τ = 1/2JCH, while the second carbon pulse returns the magnetization back to protons. By changing the phase of the first carbon pulse and that of the receiver, the relative orientation of non-13C coupled protons vs 13C coupled protons can be inverted, which leads to cancellation of non-13C coupled proton signals over two scans.

signal ultimately limits the lowest concentration of PEG, or any analyte, that can be observed using 1H NMR. The inherent proton background signal of complex samples can be significantly reduced by resorting to 13C-enriched PEG in combination with NMR pulse sequences that route the observed proton magnetization through a directly attached 13C nucleus. Well-known examples of such sequences include the heteronuclear multiple quantum coherence (HMQC) and heteronuclear single quantum coherence (HSQC) pulse schemes.18 In effect, only signals from protons directly attached to 13C survive the filter imposed by the pulse sequence. Figure 2 illustrates the HMQC pulse sequence employed here. An appropriate choice of the first delay (τ = 1/2JCH, where JCH = 144 Hz is the one-bond carbon−proton scalar coupling) transfers 1H magnetization to 13C, while the second equivalent τ delay transfers magnetization back to 1H for detection. The total transfer time of 2τ ≈ 6.94 ms associated with the 13C-filter is only a small fraction of the T2 of PEG protons, that is, T2 ∼ 600 ms for PEG5k, so signal attenuation due to T2 relaxation can be ignored. The mixing delay (δ) is set to the shortest possible time (0 ms) since we are using a one-dimensional version of HMQC. By an appropriate choice of the phase of the 13 C pulses one may invert the phase of any non-13C coupled protons such that, over a two scan sequence, the signals from the latter protons cancel out, leaving only signals from protons directly coupled to 13C. Since 13C nuclei are present at 1% natural abundance, the theoretical reduction in background 1H NMR signals is 99%. The series of 1H NMR spectra on the right in Figure 1 demonstrate the added utility of the HMQC 13C-filtered 1H NMR strategy for quantitative detection of 13C-PEG28k in a complex biological fluid, in this case goat serum. Figure 1d was obtained in the absence of added 13C-PEG28k and without 13Cfiltering. The 1H NMR background resonances, while lower in intensity compared to blood, as expected, are still intense. Figure 1e shows the 1H NMR spectrum obtained by direct detection, that is, without HMQC filtering, in the presence of a low concentration of 13C-PEG28k (serum concentration 1 μg/ mL). The PEG proton signal remains completely masked by the background resonances. However, upon employing the HMQC 13C-filtering scheme (Figure 1f), the background proton signals are virtually eliminated, with the exception of a residual water signal at ∼4.6 ppm. The latter arises from magnetization that has leaked through the filter due to rf pulse

prevent clotting. NMR spectra were acquired as described for the calibration curves.



RESULTS AND DISCUSSION Figure 1 provides a spectroscopic demonstration of the ability to detect PEGylated molecules in complex biological fluids using NMR. The series of 1H NMR spectra in the left panel were obtained from pig blood to which increasing amounts of PEG5k had been added. In the absence of added PEG (Figure 1a), blood exhibits multiple resonances in both the aliphatic (0−4 ppm) and aromatic (6−9 ppm) regions of the spectrum, arising from the various cellular constituents, proteins, and small molecules present in this complex mixture. Normally, the significant blood or serum background signal would be expected to impede the identification and quantification of any molecule present at low concentrations in such a milieu. However, the intrinsic chemical and physical properties of PEG provide an inherent signal amplification. Specifically, all monomer units of PEG are identical (excepting end units), and all four protons of the ethylene oxide repeat unit are chemically equivalent. Thus, all PEG protons appear at an identical chemical shift (3.6 ppm) in the 1H NMR spectrum and this position is conveniently separated from the dominant water resonance and from most of the aliphatic resonances, making it easy to distinguish. Furthermore, rapid rotational isomerizations about the carbon−carbon and carbon−oxygen bonds of the PEG backbone occur under most circumstances, and this produces an isotropically narrow NMR resonance, virtually regardless of the PEG size. A single 5 kDa PEG molecule contains on the order of 450 identical protons, producing a corresponding amplification of its characteristic signal, while a 100 kDa PEG molecule contains on the order of 9000 identical protons. This fortuitous combination of narrow line width, distinct chemical shift, and amplified signal intensity renders PEG visible, even from within the complex molecular milieu of blood. This is evident even for the case of a dilute PEG5k sample (50 μg/mL; Figure 1b) and abundantly so for the case of a more concentrated PEG5k sample (1000 μg/mL; Figure 1c). The spectral insets shown for the dilute PEG5k sample demonstrate that the background signal from the pure blood spectrum is small, relative to that of PEG5k or the integration control DMF. Nevertheless, this background proton E

DOI: 10.1021/acs.analchem.5b04565 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 3. Calibration curves for 1H NMR signals from (a) PEG5k in pig blood acquired via direct pulse-acquire detection, and (b) 13C-PEG28k in goat serum acquired via HMQC filtering. The slopes were 0.984 and 0.958 for PEG5k and 13C-PEG28k, respectively, while their corresponding correlation coefficients were 0.9996 and 0.9976.

Figure 4. Spectrum and a calibration curve for 1H NMR signals from PEG5k-BSA. The spectrum (a) was acquired from a blood sample collected 24 h after an intravenous injection of PEG5k-BSA in rats. The calibration curve (b) was determined in goat serum. The slope is 1.0636, while the correlation coefficient is 0.9986.

and phase imperfections and field drift. Importantly, in the region of the PEG resonance (3.6 ppm), the background signal is effectively removed and the PEG resonance at that same low concentration is readily detected. Effective NMR detection ranges and limits for PEG present in complex biological fluids were established from calibration curves. As shown in Figure 3a, direct 1H NMR detection of PEG5k in pig blood yielded a linear relationship between signal intensity and concentration over a concentration range spanning 60−4000 μg/mL. Since the ratio of the signal to background integrals was 60 for the lowest tested concentration (60 μg/mL), amounts as low as 10 μg/mL should be quantifiable using this method. For direct 1H NMR detection, the limit of detection is dictated by the relative intensities of the background versus the PEG proton signals, rather than the instrument’s sensitivity. This is evident because at this PEG concentration (60 μg/mL) the PEG proton content is relatively high (∼1.36 mM) and otherwise readily detected by the instrument. Figure 3b shows that the signal intensity from the HMQC 13C-filtered 1H NMR spectrum of 13C-PEG28k present in goat serum was linear with concentration over the range spanning 1 to 500 μg/mL. This limit of detection is an order of magnitude lower in concentration than achievable with direct

1

H NMR detection of PEG. The spectrum at the lowest concentration tested, 1 μg/mL, required ∼9 h of signal accumulation to achieve a signal-to-noise ratio (SNR) of ∼5. Thus, as opposed to the case with direct 1H NMR detection, where background signal made detection of low amounts of PEG difficult, HMQC-based 13C-filtered 1H NMR detection of 13 C-PEG is limited by the sensitivity of the spectrometer. However, as these spectra were acquired with a long recovery time (16.5 s), use of a relaxation agent could potentially decrease the experiment duration by ∼10−20-fold with little loss of resolution. Figure 4a shows a 1H NMR spectrum of PEG5k-BSA typical of those obtained during the course of blood clearance studies in live rats, to be described below (see Figure 5). Here a 50 μL aliquot of blood was withdrawn several hours after injection of PEG5k-BSA and diluted with 350 μL of D2O containing the integration standard dioxane. The PEG resonance at 3.6 ppm is quite evident, as is the fact that its line width is comparably narrow to that of free PEG shown in Figure 1 and the small molecule integration standard dioxane. Spin echo measurements show that T2 of PEG conjugated to BSA is little different from that of free PEG in solution, as detailed in Table 1, in accord with the precedent in the literature showing that PEG5k F

DOI: 10.1021/acs.analchem.5b04565 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

individual diffusion coefficients are listed in Table 1 along with the hydrodynamic diameter calculated via the Stokes−Einstein equation, as detailed in the Supporting Information. For the various free PEGs, the calculation yielded the expected dependence of the hydrodynamic diameter on the squareroot of the polymer length. 13C-PEG28k was an exception in that its diffusion coefficient was larger and its hydrodynamic size smaller than that of PEG20k. We attribute this anomaly to a greater polydispersity of this polymer synthesized “in house” relative to commercially purchased PEG. The diameter of BSA measured in this fashion increased from 8.1 nm, close to literature values,19 to 13.6 nm upon conjugation to form PEG5k-BSA. This scale of increase is consistent with the presence of 5−6 PEG5k conjugated to each BSA. Specifically, small-angle X-ray scattering (SAXS) studies show that the diameter of hemoglobin increased from 7.5 to 10.0 nm upon PEGylation with 2 PEG5k units and to 14.0 nm upon conjugation with 6 PEG5k units,20 that is, in good agreement with our diffusion results. Blood clearance profiles over the course of 24 h postinjection, as determined by the 1H NMR method, are shown in Figure 5. Fastest clearance was observed for the smaller PEG moieties, PEG20k (8.9 nm dia.) and 13C-PEG28k (8.2 nm dia.). It is likely that clearance of the smaller PEG moieties occurs via renal excretion, where the size cutoff is in the region of 6−8 nm. For example, the protein ScFv (5.3 nm) has a blood halflife of 11 min with 74% filtered into the urine, while virtually no IgG (11 nm, blood half-life of 330 min) accumulates in the urine.21,22 Although the hydrodynamic diameters of PEG20k and 13 C-PEG28k exceed somewhat the renal cutoff, PEG has high internal flexibility (relative to a globular protein), which may render it more malleable to the renal pressures responsible for clearance. Significantly slower clearance was observed for the larger moieties, PEG100k (21.7 nm dia.) and PEG5k-BSA (13.6 nm dia.), where the size exceeds by far the renal cutoff. These two larger species exhibited some scatter in the intensity levels at early time points (