Measuring Protein Concentration by Diffusion-Filtered Quantitative

Jan 4, 2019 - Poppe, L.; Jordan, J. B.; Lawson, K.; Jerums, M.; Apostol, I.; Schnier, P. D. Profiling Formulated Monoclonal Antibodies by 1H NMR ...
0 downloads 0 Views 1MB Size
Article Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/ac

Measuring Protein Concentration by Diffusion-Filtered Quantitative Nuclear Magnetic Resonance Spectroscopy Scott A. Bradley,* Wesley C. Jackson, Jr., and Patrick P. Mahoney Eli Lilly and Company, Indianapolis, Indiana 46285, United States

Downloaded via IOWA STATE UNIV on January 19, 2019 at 00:48:27 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: The concentration of macromolecules in solution is a crucial property in many areas of research, including the development and commercialization of biological therapeutics. For proteins in particular, none of the reported methods for measuring concentration detect a molecular property that is known a priori; rather, they rely on ligand binding, degradation and derivitization, or an intrinsic property that must be determined experimentally. The purpose of this report is to describe (1) a diffusionfiltered qNMR experiment (DF-qNMR) for quantitating macromolecules in complex matrices and (2) an overall method for measuring absolute protein concentration based on this DF-qNMR experiment. This method combines protein denaturation with the diffusion filter to produce clean spectra of the protein with well-resolved resonances, regardless of the matrix complexity. The concentration is then obtained by comparing the peak area of the valine/isoleucine/leucine methyl groups to an external, certified, small-molecule quantitation standard. The method, which is referred to as VILMHA (valine isoleucine leucine methyl hydrogen analysis), was tested on three proteins of various sizes. In all cases, the measured concentration was within 1.8% of the labeled value for the undiluted standard reference material evaluated. In addition, the RSD’s were less than 1.25% in all cases and less than 1% in most cases. The accuracy, precision, and ease of use make this method superior to existing absolute protein concentration methods. Furthermore, VILMHA is ideally suited to serve as the basis for converting the relative protein concentration methods into absolute methods or establishing molecular-specific parameters. Finally, DF-qNMR has the potential to quantitate other types of macromolecules (e.g., such as polymers, surfactants, etc.) in the presence of small-molecule contaminants.

K

an ion.4,5 These chromogenic methods are very sensitive, but many depend on the composition of the protein, suffer interference from the formulation component, and require calibration curves to provide the absolute concentration. A fourth approach is to degrade the protein into components that are more easily quantitated. One such method is the Kjeldahl nitrogen determination,6 which quantitates ammonia from nitrogen liberated by acid degradation of the protein. However, while it is universal and precise, it is not specific for proteins and relies on harsh conditions. Another of this type is amino acid analysis (AAA),7,8 which has been the industry standard for many years. This method uses strong acid to hydrolyze the protein to its component amino acids, which are then derivatized with a UV chromophore and quantitated. The original protein concentration is then calculated based on its known primary sequence. Unfortunately, the harsh condition of AAA creates several problems, such as the degradation of several key amino acids during the hydrolysis and derivatization steps, long run times, and high variability.9

nowing the concentration of a macromolecule in solution is critical in many situations. For example, in the pharmaceutical industry, an accurate and precise protein concentration is required for determining the biotherapeutic’s efficacy and integrating data from the various functional areas (PD/PK, ADME, toxicology, formulation, purification, clinical testing, manufacturing, etc.) into a cohesive package for regulatory submission. There are many different ways to determine the protein concentration of a solution. The most rapid and convenient way is to use the molecule’s own properties, such as the UV extinction coefficient (ε) or differential refractive index increment (dn/dc). However, these parameters are not known a priori but must be extracted from the measured response of a sample whose concentration is already known. Note that for UV, it is possible to predict ε from empirical calculations, such as that of Pace et al.,1,2 but the result is only an estimate and not the actual value. A second approach is gravimetric analysis.3 Theoretically, this should be the best approach to obtain absolute concentration. In reality, lyophilized proteins may contain a significant amount of bound water, salts and/or other formulation components, making the results inaccurate. A third approach is to measure the colored product formed when the protein binds to a dye or © XXXX American Chemical Society

Received: September 19, 2018 Accepted: January 2, 2019 Published: January 4, 2019 A

DOI: 10.1021/acs.analchem.8b04283 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

NMR Data Acquisition. All NMR experiments were conducted on an Agilent DD2 600 MHz spectrometer equipped with an Agilent 1H-19F/15N-31P PFG OneProbe. The probe temperature and pulse-field gradients were calibrated with samples of ethylene glycol and 1% H2O in D2O, respectively. The spectra were acquired at 30.0 °C. The probe was carefully tuned and matched manually, and the pulse width was measured on each sample prior to data acquisition. For the external quantitation standard, three spectra were acquired using the standard one-pulse experiment with a 20-ppm spectral width, a 1.363-s acquisition time, a 60-s relaxation delay, and 8 scans. The sample was ejected, reinserted, tuned, locked, shimmed, and calibrated between acquisitions. For the DF-qNMR experiments, typical acquisition parameters included a 20-ppm spectral width, a 1.363-s acquisition time, a 30-s relaxation delay, a 150-ms diffusion delay, 1.4-ms gradient pulses that were 0.569 T/m strong (92% of maximum), and a 1-ms gradient stabilization delay. For measuring the diffusion coefficient, typical acquisition parameters included a 1-s relaxation delay, a 200-ms diffusion delay, a 2.0-ms diffusion gradient length, a 0.5-ms gradient stabilization delay, and seven values of the gradient strength ranging from 0.228 to 0.569 T/m (ca. 37−92% of maximum) with logarithmic spacing. The T1 relaxation times were measured with the bppste pulse sequence using a 20-ppm spectral width, a 1.363-s acquisition time, a 3.637-s relaxation delay, a 2.0-ms diffusion gradient length, and a 0.5-ms gradient stabilization delay. Five pairs of the diffusion delay and gradient strength(100 ms, 0.569 T/m); (203 ms, 0.400 T/ m); (408 ms, 0.281 T/m); (804 ms, 0.200 T/m); and (1202 ms, 0.164 T/m)were used. The T2 relaxation time was also measured with the bppste pulse sequence using a 20-ppm spectral width, a 1.363-s acquisition time, a 1-s relaxation delay, and a 1.4-ms diffusion gradient length. Six sets of the gradient stabilization delay, diffusion delay, and gradient strength(1 ms, 150.001 ms, 0.570 T/m); (2 ms, 150.002 ms, 0.571 T/m); (4 ms, 150.004 ms, 0.573 T/m); (8 ms, 150.008 ms, 0.577 T/ m); (12 ms, 150.012 ms, 0.581 T/m); and (16 ms, 150.016 ms, 0.585 T/m)were used. (See the Supporting Information for additional details.) The number of scans was 64 for all experiments unless otherwise stated. NMR Data Analysis. The external standard, DF-qNMR, T1, and T2 data were processed in MNova version 11.0 (Mestrelab Research, S. L., Santiago de Compostela, Spain), while the diffusion data were processed and analyzed with DOSYToolbox version 2.5.18 In all cases, the FIDs were zerofilled one time and multiplied by an exponential window function of 2.93 Hz prior to Fourier transform. The peak areas of maleic acid in the external standard and the leucine, isoleucine, and valine methyls (1.0−1.4) in the protein samples were obtained by deconvolution using the line fitting routine in MNova or CRAFT.19 The area of the maleic acid peak from the three spectra were averaged, and the average value was used in subsequent concentration calculations (ARS). The values of T1 and T2 were calculated from the bppste pulse sequence by regression analysis using MATLAB 2016b (MathWorks, Inc., Natick, Ma), as described in the Supporting Information. The diffusion coefficients were obtained using the DECRA algorithm available in DOSYToolbox. Density Measurements. Sample densities were analyzed using a Mettler Toledo DE-40 Density Meter.

It would be beneficial to have an accurate absolute protein concentration method that is indifferent to the protein structure and formulation components, does not rely on molecular interactions or calibration curves to provide absolute concentrations, and avoids overly harsh sample preparation. NMR is an intrinsically quantitative spectroscopic technique.10,11 In addition to being a proven method for measuring the purity and potency of small-molecule solids, quantitative NMR (qNMR) has been frequently used in a number of applications to measure the concentration of compounds that cannot be isolated as pure solids.10−16 However, using qNMR to measure the concentration of large proteins in these types of complex matrices has been disregarded because of two challenges. First, the NMR spectra are dominated by the intense peaks from the various salts, buffers, surfactants, tonicity agents, and water. Second, the high-order structure of most proteins and antibodies creates unique magnetic environments around the hydrogen atoms that cause larger 1 H chemical shift windows for a given amino acid. Together with inherently broader line widths for protein resonances, it is difficult to have sufficient resolution for accurate peak identification and integration.17 Herein, we present a diffusion-filtered qNMR experiment (“DF-qNMR”), demonstrate its effectiveness at dampening the resonances from the matrix components, and describe the calculations required to determine absolute concentration from the resulting spectra. We also present a method to determine the absolute protein concentration in complex formulations, which uses the peak area of the valine/isoleucine/leucine methyl groups from the DF-qNMR spectrum of the denatured protein. We refer to this method as VILMHA (valine isoleucine leucine methyl hydrogen analysis). The accuracy and precision of the method will be demonstrated on three proteins of various sizes and in different formulations.



EXPERIMENTAL SECTION Materials. Maleic acid (TraceCERT certified qNMR reference material, 99.94%) and D2O were purchased from Sigma-Aldrich. Guanidine deuterochloride (CD6N3Cl, 98%) was purchased from Cambridge Isotope Laboratories. NIST RM 8761 (10 g/L antibody, 12.5 mM L-histidine HCl, pH 6.0) and 927e (67.38 g/L, 20 mM NaCl pH adjusted to 6.5−6.8 with 1.0 mol/L NaOH) were purchased from the National Institute of Standards and Technology. Precision 4-mm NMR tubes (Wilmad 435) were purchased from Wilmad Labglass. NMR Sample Preparation. The external quantitation reference standard was prepared by placing 11.183 mg of maleic acid into a 5-mL volumetric flask and dissolving with D2O. An aliquot was transferred to a Wilmad 435 precision 4mm NMR tube such that the sample height was 40 mm. The protein test samples were prepared as follows. Approximately 0.6 g of guanidinium chloride-d6 was weighed and transferred to a 1-mL volumetric flask. The flask was placed onto a balance, the balance was tared, and 400 uL of sample was added using an Eppendorf digital pipette. The weight was recorded for subsequent calculations. The solution was then gently sonicated and vortexed until all solids were dissolved. (Note that 400 uL of protein solution is near the maximum amount possible because of the significant volume expansion from denaturation.) D2O was added to achieve the final volume of 1 mL. An aliquot was transferred to a Wilmad 435 precision 4 mm NMR tube such that the sample height in the tube was 40 mm. B

DOI: 10.1021/acs.analchem.8b04283 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry



RESULTS AND DISCUSSION The 1H 1D NMR spectrum of the NIST antibody (RM 8761) is shown in Figure 1A,B. The NMR sample was prepared by

provides a means to obtain the desired antibody spectrum but exposes the next significant challenge for quantitation: the width of the protein peaks. The line width that is characteristic of protein NMR peaks is primarily due to the higher-order structure (HOS) of the protein. A single type of amino acid, which would otherwise have very similar chemical shifts, shows a wide chemical shift distribution due to the unique magnetic environments resulting from the secondary, tertiary, and quaternary structure of the protein. The solution employed in this method is to remove the HOS by denaturing the protein. This is a reasonable sacrifice since (1) the purpose of the experiment is only quantitation and not the protein structure or function, (2) the method is employed on a small sampling of the material, and (3) denaturation can be done easily during NMR sample preparation. Figure 1D shows the diffusion-filtered spectrum of the NIST antibody denatured with 6 M guanidinium chloride-d6. This sample was prepared by adding ∼0.4 mL of the purchased antibody sample gravimetrically to a 1 mL volumetric flask containing enough solid guanidinium chloride-d6 to make a 6 M solution and then diluting to 1 mL with D2O. The observed decrease in line width (and the subsequent increase in signal-to-noise) is considerable. The group of peaks between 1.0 and 1.4 ppm, which correspond to the methyl groups of isoleucine, leucine, and valine, has near baseline resolution. This, along with its higher intensity, makes this group of peaks the best choice to be integrated for quantitative analysis. Deconvolution routines can be used to ensure the entire area is captured. Calculating the concentration of the solution from the aforementioned protein peak in the qNMR spectrum requires a reference standard of known concentration, as described by the following equation: c P = c RS

Figure 1. 1H NMR spectra of NIST mAb (A) in native form with a standard one-pulse sequence, (B) same as A with vertical scale 5000x, (C) native with a diffusion-filter, and (D) denatured with a diffusion filter. Peaks marked with an asterisk are from the histidine buffer. The number of scans was 1024 for A-C and 64 for D.

APHRS ARSHP

(1)

where cP and cRS are the molar concentrations of the of the protein and reference standard, AP and ARS are the areas for the protein and reference standard peaks, and HP and HRS are the number of protons contributing to each peak. For this method, an external reference standard was chosen because it avoids potential interactions between the protein and internal standard, as well as overlap of the peaks in the NMR spectrum. Several external standard methods have been reported.17,25−27 One that works particularly well is the PULCON technique (e.g., pulse length-base concentration determination).17 This technique correlates the absolute areas from two individual spectra, one of the reference standard and the other of the analyte, even if the solution conditions and experimental parameters are different. This means that the external reference standard need not have similar characteristics to the analyte of interest or even be a protein. Any pure compound of any size will work. Because small-molecule certified quantitation standards are available from several sources, solutions of known concentration can be prepared very accurately. Furthermore, rather than needing a standard for every amino acid, as is the case for AAA, this method only requires a single standard. This minimizes a source of uncertainty that is common in available protein concentration measurements. While the diffusion filter is effective at isolating the protein signals from the matrix, it introduces one complication for

adding 10% D2O to the purchased solution. This spectrum illustrates the challenges of quantitating large, formulated proteins by NMR. To begin with, the water signal is so intense and broad that it obscures the signals of the antibody. Depending on how the protein is formulated, the same may be true of the excipient peaks, such as the histidine signals observed in this spectrum. To get a viable spectrum of the protein, these dominant peaks must be eliminated. Of all the multifrequency suppression routines that have been reported,20 one technique that is ideal for this application is a diffusion filter,21 where peaks are attenuated based on the diffusion coefficients, and hence size, of the corresponding molecules. This approach has been used recently to remove residual solvent signals from the NMR spectra of organic compounds22 as well as those of matrix components from the spectra of antibody solutions.23 Figure 1C shows the 1H NMR spectrum acquired on the same sample as Figure 1A but with the bipolarpulse pair stimulated echo (bppste) diffusion pulse sequence24 as a diffusion filter. As can be seen, the diffusion filter is very effective at removing the unwanted signals of the formulation without introducing baseline artifacts or phase distortions. This C

DOI: 10.1021/acs.analchem.8b04283 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

has been extensively evaluated and serves as a good benchmark. Three independent replicates were prepared and analyzed. The DF-qNMR spectrum of one replicate is shown in Figure 1D. Ample signal-to-noise was obtained on this 10 g/ L protein formulation using only 64 scans and a 37 min total acquisition time. The group of signals for the methyl groups of I/L/V in the protein, which corresponds to 1476 protons, has near baseline resolution. These results enable accurate integration for quantitation. T1, T2, and D were measured for this peak group with the bppste pulse sequence (as described in the Supporting Information) and found to be 0.678 s, 0.0571 s, and 1.73 × 10−11 m2/s, respectively. Note that the measured T1 and T2 are averages since they were determined from the signal of multiple overlapping methyl groups. However, this is sufficient for the intended purpose, which is to account for attenuation of this signal during the delays of the pulse sequence where the magnetization is stored along the z-axis (T1) or x,y-axes (T2). Note also that ignoring this attenuation can have significant impact on the accuracy of the results. For the pulse sequence parameters used here, the uncorrected value was 30% lower than expected. The total experiment time for all four NMR experiments on one replicate was 110 min. The concentration of each replicate was then calculated from eq 3 using a dilution factor based on the weight of the aliquot and the measured sample density. The average concentration was determined to be 10.02 g/L with an RSD of 0.55%. This compares favorably with the labeled value of 10 g/L. Furthermore, the precision of the three replicates was better than those typically obtained by AAA. For a more robust test of accuracy, precision, and linearity, the method was evaluated on NIST bovine serum albumin (BSA, 66 398 Da) standard reference material 927e as received (67.38 g/L ± 1.38 g/L by AAA) and gravimetrically diluted to four additional protein concentrations (Table 1, samples B1−

qNMR experiments: the measured NMR peak areas are no longer inherently quantitative. Routine qNMR experiments use a simple NMR pulse sequence, which consists of three basic elements: a nuclear-relaxation delay, one radio frequency pulse, and a data-acquisition period. When a sufficiently long nuclear relaxation delay is used, the resulting NMR peak area is accurately and reproducibly proportional to the number of observed nuclei; thus, the experiment is quantitative. All other NMR pulse sequences, including the diffusion sequence employed here, require multiple radiofrequency pulses separated by various delays. Consequently, the resulting spectra are no longer intrinsically quantitative due to factors that attenuate the peaks from the equilibrium value. However, if those factors are known, the extent of attenuation can be calculated and the diffusion filter can be made quantitative. It can be shown that the peak area resulting from the bppste pulse sequence with square gradient pulses, A, is proportional to the equilibrium peak area, Ao, according to this specific Stejskal−Tanner equation24,28 A=

Ao −τ1/ T1 −τ2 / T2 −Dγ 2g 2δ 2(Δ− δ − τ3 ) 3 2 ] e e [e 2

(2)

where τ1, τ2, and τ3 are delay times between various pulses in the sequence; γ is the magnetogyric ratio; g and δ are the strength and length of the pulse-field gradients, respectively; Δ is the diffusion delay; T1 is the longitudinal nuclear relaxation time; T2 is the spin−spin nuclear relaxation time; and D is the translation diffusion coefficient of the molecule. In effect, the first exponential term in the equation describes signal attenuation due to T1 nuclear relaxation, the second due to T2 nuclear relaxation, and the third due to molecular diffusion. Of all the terms in the equation, these three quantities (T1, T2, and D) are the only unknowns; all others are pulse sequence/ instrument parameters or a physical constant (γ). Therefore, by measuring these three values for a given sample, the equilibrium peak areas can be calculated, and the diffusion filter becomes quantitative.29 Furthermore, they can be conveniently measured with the bppste pulse sequence itself by an appropriate choice of acquisition parameters (Supporting Information).23 Altogether, the complete equation for the molar concentration of the protein from the DF-qNMR experiment is given by

Table 1. Results for the NIST BSA Samples sample

gravimetric conc. (g/L)a

VILMHA conc. (g/L)

n

RSD

B5 B4 B3 B2 B1

67.38 52.82 38.11 23.91 9.525

66.18 52.50 37.53 23.90 9.297

3 1 6 1 3

0.82% 0.81% 0.61%

percent difference −1.78% −0.60% −1.54% −0.04% −2.39%

a

(3)

The reported expanded uncertainty for the reference material (B5) is ±1.38 g/L. The values for the other samples are likely to be higher due to the dilutions.

The factor, fq, contains additional parameters necessary for the PULCON algorithm: fq = (TP pwP SRS)/(TRS pwRS SP), where T, pw, and S are the temperature, pulse width, and number of scans, respectively, used to acquire the spectra of the protein and reference standard. The last term, fd, accounts for dilution of the original sample to the 6 M guanidinium chloride NMR sample. Since all NMR samples described here were prepared using the aliquot weight (wtP) and measured density (ρP) diluted to a final volume of 1 mL, fd = (1 mL × ρP)/wtP. Finally, if desired, the concentration can be converted to the more common units of g/L by multiplying the calculated concentration by the molecular weight of the protein. To test the accuracy and precision of the method, the antibody concentration of the aforementioned NIST RM 8761 sample was determined. This is a certified reference material of a humanized IgG1κ monoclonal antibody (148 038 Da) that

B4). Fourteen total samples were analyzed: six replicates at the middle concentration (B3), three replicates at the lowest (B1) and highest (B5) concentrations, and single replicates at the intermediate concentrations (B2 and B4). The NMR sample for each replicate was prepared as described above. The resulting 14 DF-qNMR spectra are shown superimposed in Figure 2. Again, the quality of the spectra for this large protein is excellent: the peak from water has been nearly eliminated, the baseline is unperturbed, and the group of signals for the methyl groups of I/L/V in the protein has near baseline resolution. The concentration for each sample was calculated from eq 3. The results are reported in Table 1. For sample B5 (the denatured NMR sample prepared directly from undiluted standard), the value measured by VILMHA compared favorably to the labeled concentration (67.38 ± 1.38 g/L).

c P = c RS

APHRS τ1/ T1 τ2 / T2 Dγ 2g 2δ 2(Δ− δ − τ3 ) 3 2 ]f f [2e e e q d ARSHP

D

DOI: 10.1021/acs.analchem.8b04283 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. DF-qNMR spectra of the 14 samples of denatured NIST BSA.

More significantly, the RSD was less than 1% for all three sets of replicates. A plot of the VILMHA concentrations vs the gravimetric concentrations yielded a regression line with R2 = 0.9998, indicating excellent linearity across the concentration range examined (Supporting Information). The same evaluation was performed on a third protein, a proprietary bispecific antibody with a molecular weight of approximately 200 000 Da. While it lacks the independent traceability of a purchased standard, it represents a biopharmaceutical product with a very high molecular weight. As before, five different concentrations and 14 total samples were analyzed. Figure 3 shows a portion of the resulting DFqNMR spectra. Even though the molecular weight is over three times that of BSA, the spectra are strikingly similar in overall appearance and quality. While the group of peaks for the methyl groups of I/L/V in the protein, which represents over 300 methyls and nearly 2000 total protons, is slightly broader, it still has near baseline resolution. The results are summarized in Table 2. The accuracy compared favorably to the UV method using the theoretical extinction coefficient, and the precision were found to be better than existing methodologies. A plot of the VILMHA concentrations vs the gravimetric concentrations yielded a regression line with R2 = 0.9966, demonstrating linearity across the concentration range examined (Supporting Information).

Figure 3. DF-qNMR spectra of the 14 samples of the denatured bispecific antibody. The peak for residual water is marked with an asterisk.

Table 2. Results for the Bispecific Antibody



CONCLUSION A new qNMR method was developed to measure the absolute concentration of proteins. Sample preparation and data acquisition were easier, quicker, and safer than the current AAA method. Excellent NMR spectra of the formulated proteins were obtained by eliminating the unwanted signals of the matrix with a quantitative diffusion filter and reducing the

sample

UV conc. (g/L)a

VILMHA conc. (g/L)

n

RSD

B5 B4 B3 B2 B1

74.01 55.02 39.67 24.61 9.277

78.37 54.43 41.62 24.59 9.427

3 1 6 1 3

0.25%

a

E

1.25% 0.85%

percent difference 5.89% −1.07% 4.92% −0.09% 1.61%

Determined by UV using the theoretical extinction coefficient.

DOI: 10.1021/acs.analchem.8b04283 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(16) Pauli, G. F.; Jaki, B. U.; Lankin, D. C. J. Nat. Prod. 2007, 70, 589−595. (17) Wider, G.; Dreier, L. J. Am. Chem. Soc. 2006, 128, 2571−2576. (18) Nilsson, M. J. Magn. Reson. 2009, 200, 296−302. (19) Krishnamurthy, K. Magn. Reson. Chem. 2013, 51, 821−829. (20) McKay, R. T. Annu. Rep. NMR Spectrosc. 2009, 66, 33−76. (21) Stilbs, P. Prog. Nucl. Magn. Reson. Spectrosc. 1987, 19, 1−45. (22) Esturau, N.; Espinosa, J. F. J. Org. Chem. 2006, 71, 4103−4110. (23) Poppe, L.; Jordan, J. B.; Lawson, K.; Jerums, M.; Apostol, I.; Schnier, P. D. Anal. Chem. 2013, 85, 9623−9629. (24) Johnson, C. S., Jr. Prog. Nucl. Magn. Reson. Spectrosc. 1999, 34, 203−256. (25) Akoka, S.; Barantin, L.; Trierweiler, M. Anal. Chem. 1999, 71, 2554−2557. (26) Burton, I. W.; Quilliam, M. A.; Walter, J. A. Anal. Chem. 2005, 77, 3123−3131. (27) Farrant, R. D.; Hollerton, J. C.; Lynn, S. M.; Provera, S.; Sidebottom, P. J.; Upton, R. J. Magn. Reson. Chem. 2010, 48, 753− 762. (28) Sinnaeve, D. Concepts Magn. Reson., Part A 2012, 40A, 39−65. (29) Barrère, C.; Thureau, P.; Thévand, A.; Viel, S. J. Magn. Reson. 2012, 216, 201−208.

line widths of the protein by denaturation. The linearity, precision, and accuracy of the resulting concentrations were found to be suitable for a variety of proteins and formulations. Overall, VILMHA is a reliable option for measuring the absolute protein concentrations, including peptides that lack chromophoric amino acids and are difficult to monitor by UV. Furthermore, VILMHA is ideally suited to serve as the basis for measuring other intrinsic parameters of the protein or converting any of the current relative concentration methods into an absolute method. Finally, the broader DF-qNMR technique enables quantitation of other types of macromolecules (such as polymers, surfactants, carbohydrates, etc.) in the presence of smaller molecules.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.8b04283. Example of diffusion measurement and data (Figure S1); measuring T1 and T2 with the bppste pulse sequence on Agilent NMR spectrometers (Figures S2 and S3); example of integration the protein peak by line fitting (Figure S4); linearity plots for NIST BSA and the bispecific antibody (Figures S5 and S6) (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: 1-317-651-9685. ORCID

Scott A. Bradley: 0000-0003-3419-2445 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank Will Weiss, Brandon Doyle, Elisabeth Krug, Eric Adamec, Charles Mitchell, and Chad Hadden for advice and technical guidance.



REFERENCES

(1) Edelhoch, H. Biochemistry 1967, 6, 1948−1954. (2) Pace, C. N.; Vajdos, F.; Fee, L.; Grimsley, G.; Gray, T. Protein Sci. 1995, 4, 2411−2423. (3) Nozaki, Y. Arch. Biochem. Biophys. 1986, 249, 437−446. (4) Bradford, M. M. Anal. Biochem. 1976, 72, 248−254. (5) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 193, 265−275. (6) Jaenicke, L. Anal. Biochem. 1974, 61, 623−627. (7) Spackman, D. H.; Stein, W. H.; Moore, S. Anal. Chem. 1958, 30, 1190−1206. (8) Benson, A. M.; Suruda, A. J.; Talalay, P. J. Biol. Chem. 1975, 250, 276−280. (9) Sittampalam, G. S.; Ellis, R. M.; Miner, D. J.; Rickard, E. C.; Clodfelter, D. K. Journal of the Association of Official Analytical Chemists 1988, 71, 833−838. (10) Malz, F.; Jancke, H. J. Pharm. Biomed. Anal. 2005, 38, 813−823. (11) Bharti, S. K.; Roy, R. TrAC, Trends Anal. Chem. 2012, 35, 5− 26. (12) Webster, G.; Kumar, S. Anal. Chem. 2014, 86, 11474−11480. (13) Mahajan, S.; Singh, I. P. Magn. Reson. Chem. 2013, 51, 76−81. (14) Pauli, G. F.; Jaki, B. U.; Lankin, D. C. J. Nat. Prod. 2005, 68, 133−149. (15) Liang, X.; Du, L.; Su, F.; Parekh, H. S.; Su, W. Magn. Reson. Chem. 2014, 52, 178−182. F

DOI: 10.1021/acs.analchem.8b04283 Anal. Chem. XXXX, XXX, XXX−XXX