Integrated Targeted Quantitation Method for Insulin and its Therapeutic Analogs

Transcription

1 Integrated Targeted Quantitation Method for Insulin and its Therapeutic Analogs Eric Niederkofler, 1 Dobrin Nedelkov, 1 Urban Kiernan, 1 David Phillips, 1 Kemmons Tubbs, 1 Scott Peterman, 2 Bryan Krastins, 2 Amol Prakash, 2 and Mary Lopez 2 1 Thermo Fisher Scientific, Tempe, AZ; 2 Thermo Fisher Scientific BRIMS, Cambridge, MA

2 Overview Purpose: Perform simultaneous qualitative measurements and quantitation on endogenous insulin and/or therapeutic analogs at biological levels for research. Methods: Incorporate pan-insulin Ab in the Thermo Scientific MSIA (Mass Spectrometric Immunoassay) Tips for increased extraction efficiency of all insulin variants that are detected, verified, and quantified using HR/AM MS and MS/MS data on the Thermo Scientific Q Exactive mass spectrometer. Results: Quantitation ranges reached nm in plasma for all variants used in the experiment with linear regressions of 0.99 or better. In addition, robust qual/quan results were observed for multiple insulin variants spiked at different levels. Introduction The need to detect and quantify insulin and its therapeutic analogs has become paramount for many different research assays 1. Insulin is typically present at sub ng/ml in the presence of complex biological matrices requiring extraction/enrichment protocols to be used prior to LC-MS detection and quantitation. In addition to endogenous insulin quantification, variants are also used to stimulate the same response and need to be quantified. Variants contain slight sequence variations to effect bioavailability and are generally administered at sub ng/ml levels. To reduce sample handling bias, a universal extraction method is required to facilitate simultaneous insulin variant extraction for targeted quantitation. In addition, the LC-MS detection method must be amenable to detection and quantification of known and unknown variants. Methods Sample Preparation All samples were prepared from a stock solution of plasma. To each well a 500 µl aliquot of the plasma was added as well as 0.05 nm porcine insulin and used as an internal standard. Three different sets of samples were prepared in the wells. The first set had individual insulin variants spiked covering a range of to 0.96 nm increasing in 2-fold steps. The second set of samples had one insulin variant spiked covering the same concentration range as that in sample set 1 except Humulin S was spiked in at a constant concentration of 0.06 nm. The last set of samples spiked two different insulin variants over the expressed concentration range with Humulin S spiked in at a constant concentration of 0.06 nm. Each sample was extracted using a MSIA Thermo Scientific D.A.R.T. (Disposable Automated Research Tips) loaded with 3 µg of pan-insulin Ab in an automated method using the Thermo Scientific Versette Automated Liquid Handler 2. Following insulin extraction, washing, and elution into a new well, the samples were dried down and then reconstituted in a µl solution of 75:25:0.2% water/mecn/formic acid with 15 mg/ml ACTH Liquid Chromatography (or more generically Separations) An Thermo Scientific Dionex UltiMate 3000 RSLC system was used for all experiments and µl of each sample was separated on a x 1 mm Thermo Scientific ProSwift RP-4H 1 x 250 mm monolithic column using a linear gradient (10-50% in 10 minutes) comprised of A) 0.1% formic acid in water and B) 0.1% formic acid in MeCN. The column was heated to a temperature of 50 ºC. Mass Spectrometry All experiments were acquired using a Q Exactive mass spectrometer operated in data-dependent/dynamic exclusion. A resolution setting of 70,000 (@m/z 0) was used for full scan MS and 17,500 for MS/MS events. Full scan MS data was acquired using a mass range of 0-00 Da and a targeted inclusion list was used to trigger all data dependent events. Data Analysis All data was processed using Thermo Scientific Pinpoint 1.3 software. HR/AM MS data extraction was used for quantitation. To provide additional levels of qualitative analysis, the three most abundant precursor charge states per insulin variant were used as well as the six most abundant isotopes per charge state. A mass tolerance of ±5 ppm was used for all data extraction. Qualitative scoring was based on mass error, precursor charge state distribution, and isotopic overlap as well as corresponding LC elution peak profiles measured for each sample. Product ion data was used for sequence verification. The measured AUC values for porcine insulin was used as an internal standard for all samples. 2 Integrated Targeted Quantitation Method for Insulin and its Therapeutic Analogs

3 Results The protocol for targeted detection and quantification of insulin and different insulin sequence variants must have specific attributes to be effective. The sensitivity and selectivity of extraction and detection methods must reach biological levels as well as provide qualitative measurements per target. A useful internal standard was included to normalize the entire method from the sample preparation, LC-MS analysis, and data processing. Lastly, the protocol must be effective for most insulin variants to reduce cost and complexity for the workflow. Our workflow has been shown to reach the required biological levels, facilitate a lowcost internal standard in porcine insulin, and automate the workflow to expedite sample analysis and data processing. The key aspect is based on effective targeted extraction using the Ab coated MSIA tips. Figure 1 shows the automated steps to first bind the insulin variants, wash off background compounds, and elution into a new plate. Once the extraction was performed, the plate was then prepped for LC-MS analysis. This process eliminates the two steps previously reported while increasing the detection/quantitative capabilities using the Q Exactive mass spectrometer. FIGURE 1. Targeted extraction process using covalently bounded pan-insulin Ab to MSIA D.A.R.T tips. All samples were processed using the same protocol. Following automated extraction, washing, and elution, the samples were dried down prior to being reconstituted in an LC-MS solvent composition. The subsequent LC-MS detection using HR/AM MS data enabled sufficient selectivity to distinguish insulin variants from the background signal using multiple precursor charge states and isotopes. The data extraction approach as shown in Figure 2 demonstrates multliple verification attributes from the LC and MS profiles. Data dependent MS/MS acquisition can also be used for specific variant determination as well. (data not shown) Decoupling the quantitative method (MS data) from sequence confirmation (MS/MS) enables the method to probe not only for known variants, but to perform significant postacquisition processing as new variants become known, provided the b-chain epitope region remains consistent. Figure 3 shows the comparative extraction and detection efficiency of the workflow across five different insulin variants. FIGURE 2. Targeted data extraction approach in the Pinpoint 1.3 software based on HR/AM MS data. Data from each targeted insulin variant was extracted based on isotopic m/z values from three precursor charge states. Integrated AUC values from each isotope was co-added to generate the reported values. In addition, qualitative analysis was performed to score each insulin variant based on 2A) comparative peak profiles (peak stop and stop, apex, and tailing factors) as well as 2b) isotopic distribution overlap. 2a 2b Thermo Scientifi c Poster Note PN ASMS13_M503_ENiederkofl er_e 07/13S 3

4 FIGURE 3. Comparative analysis of insulin variant extraction using a common workflow, including the same tips, LC separation, MS data acquisition, and data processing. Each sample was prepared by spiking 0.24 nm of each variant in different wells. The measured results are listed in each figure as well as the results for porcine (Figure 3e). The Pinpoint processing method included precursor m/z values for each variant. 3a 3b 3c Humulin S Lantus Bovine 69:: e ppm 91:: e ppm 91:: e ppm Hum Lan Bov Nov 3d Novorapid 63:: e ppm Hum Lan Bov Nov Hum Lan Bov Nov 3e 75:: e ppm Hum Lan Bov Nov Hum Lan Bov Nov FIGURE 4. Targeted quantitation curve for Humulin S. The measured AUC values were summed from 18 isotopic m/z value across three charge states. R 2 = y = x FIGURE 5. Qualitative output from Pinpoint software to evaluate 5a) precursor charge state and 5b) +5 isotopic distribution for each spiked Humulin S levels in plasma. Humulin S Amount Spiked into Plasma Mono A+5 A+4 A+1 A+3 A Amount Spiked into Plasma Mono A+5 A+4 A+1 A+3 A+2 4 Integrated Targeted Quantitation Method for Insulin and its Therapeutic Analogs

5 FIGURE 6. Normalized quantitative curve for Humulin S in plasma. The measured porcine response was used to normalize each level. R 2 = y = x FIGURE 7. Normalized quantitative curves for bovine and Lantus insulin variants. Each variant was spiked into the plasma separately with a constant amount of porcine in all samples. The relative curves are reflective of the measured response shown in Figure 3. FIGURE 8. LC-MS data analysis of sample processing of four insulin variants spiked into plasma. The four samples are Lantus and Glulisine spiked at 0.48 nm, Humulin S (0.06 nm), and porcine as the internal standard. The resulting full scan spectrum was averaged across the three co-eluting variants. Lantus elutes 0.5 minutes prior to the three displayed insulin variants. Relative Abundance Humulin S 0.06 nm 5.5E nm 4.8E Glulisine 0.48 nm 2.2E Time (min) Relative Abundance Humulin S m/z Glulisine z= m/z Thermo Scientifi c Poster Note PN ASMS13_M503_ENiederkofl er_e 07/13S 5

6 A secondary test was performed to evaluate the effects of multiple insulin variants spiked at different levels on targeted extraction and detection. Figure 8 shows the resulting full scan MS to demonstrate the Q Exactive data acquisition of the different insulin variants. The mass spectrum shows matrix from the MSIA elution solvents that formed predominantly singly charged ions compared to the targeted insulin variants at which are ca. 2-5% of the total signal and the resolution-facilitated peak detection and extraction. Figure 9 shows the quantitation for two different insulin variants spiked over same dynamic range as well as the porcine and Humulin S variants spiked at a constant level. insulin was used as the internal standard. Despite the difference in measured signal, each variant was detected at the lower levels and the resulting linear regression was 0.99 or better. In addition, the amount of Humulin S could be determined based on the individual quan curve presented in Figure 6. Using the y value of from Figure 9 and the linear equation in Figure 6, the calculated amount was nm compared to the predicted amount of 0.06 nm. FIGURE 9. Comparative quantitation curves for Lantus and Glulisine that were spiked into plasma at different levels as well as Humulin S which was spiked into each sample at a constant amount of 0.06 nm to replicate endogenous insulin. All AUC values were normalized to the porcine AUC response. Conclusion The targeted workflow successfully demonstrated selective and sensitive insulin variant extraction, LC separation, and quantitation using HR/AM MS for research. The pan insulin Ab was sensitive for all six insulin variants used in the study. The automated sample extraction utilized one step as opposed to multiple enrichment/extraction steps. Detection and quantitation ranges reached were at nm in 0.5 L of plasma. The MSIA D.A.R.T. tips showed equivalent extraction and quantitative efficiency for singly or multiply spiked insulin variants at different concentration ranges. Decoupling data used for quantitative and qualitative analysis facilitates reprocessing for potential unknown insulin variants. The Pinpoint 1.3 software provides automated data extraction, verification, and quantification for all insulin variants. References 1. Thevis, M, Thomas, A., Schänzer, W. Mass Spectom. Reviews 08, 27(1), Nelson, R. W., Krone, J., R., Bieber, A. L., Williams, P. Anal. Chem. 1995, 67, Acknowledgements The authors would like to thank Dr. Stephan Morely from the Sheffield Hospital, UK for the donation of the insulin variants used in the study. 6 Integrated Targeted Quantitation Method for Insulin and its Therapeutic Analogs

7 13 Thermo Fisher Scientific Inc. All rights reserved. ISO is a trademark of the International Standards Organization. Humulin is a registered trademark of Eli Lilly and Company. All other trademarks are the property of Thermo Fisher Scientific, Inc. and its subsidiaries. Specifications, terms and pricing are subject to change. Not all products are available in all countries. Please consult your local sales representative for details. Thermo Fisher Scientific, San Jose, CA USA is ISO 9001:08 Certified. Africa-Other Australia Austria Belgium Canada China Denmark Europe-Other Finland/Norway/Sweden France Germany India Italy Japan Latin America Middle East Netherlands New Zealand Russia/CIS South Africa Spain Switzerland UK USA ASMS13_M503_ENiederkofler_E 07/13S