R eview For reprint orders, please contact [email protected]
Sensitive glucagon quantification by immunochemical and LC–MS/MS methods The peptide hormone glucagon plays an important role in homeostasis of glucose concentrations in the blood. Its biological importance is evidenced through the conservation of its peptide sequence between species. Reliable assays for glucagon in biological samples are important for gaining a better understanding of the pathology and treatment of diabetes. Numerous assays are available for the analysis of glucagon in biological samples, the majority of which employ an immunochemical approach and have been available for many years. However, recent advances in MS instrumentation and the amenability of glucagon for analysis by LC–MS/MS has brought these new methods to the forefront. Concentrations of glucagon determined from different methods are not always consistent and this review provides suggestions of how to improve the reliability of methods for glucagon analysis.
Diabetes is a group of metabolic diseases characterized by a high level of glucose in the blood. A significant research effort is currently directed towards understanding and treating diabetes, which was estimated to cost the USA approximately US$175 billion for direct and indirect healthcare in 2007 [1]. In 2012, nearly 10,000 articles with diabetes in the title (PubMed search) were published. Insulin and glucagon are peptide hormones that are responsible for the maintenance of normal glucose levels in blood. Insulin acts in response to an elevated blood glucose concentration by increasing the uptake of glucose by cells, while the primary role of glucagon is to respond to hypoglycemia by increasing glycogenolysis and gluconeogenesis in the liver in order to increase the blood glucose concentration [2]. Elevated glucagon levels are suggested to contribute to the severity of hyperglycemia in diabetic patients [3]. As a therapeutic agent, glucagon may be administered as an emergency treatment to combat severe hypoglycemia associated with the use of insulin and some other medications for diabetic patients. Glucagon (a peptide chain of 29 amino acids) and two other peptide hormones, GLP-1 and GLP-2, are encoded by a single proglucagon gene in humans. Tissue-specific, post-translational processing of the proglucagon gene product, from the transcript, results in the secretion of glucagon by pancreatic a-cells, whereas intestinal b-cells produce GLP-1 and GLP-2, as reviewed by Pocai [4]. Proglucagon processing in the intestine also results in two other peptides (enteroglucagon) that contain the
glucagon sequence – oxyntomodulin and glicentin (Figure 1). The peptide sequence of glucagon is highly conserved between mammalian species, which reinforces the notion of its important biological functions. Glucagon receptors are not only expressed in the liver, but also in the kidneys, pancreas, intestines, brain and adipose tissues, as reviewed by Authier and Desbuquois [5]. Experimental data suggest that glucagon performs many other biological functions, in addition to acting as an antihypoglycemic agent, and plays an important role in stress response and energy metabolism [6,7]. Changes in glucagon concentrations as a result of dietary, metabolic and pharmaceutical interventions can provide insight into the epidemiology of diabetes and the efficacy of potential treatments. Unraveling the multifaceted functionality of glucagon strongly relies on the availability of accurate, precise and robust bioa nalytical methods. In addition, sensitivity at a low pg/ml level is required to analyze endogenous glucagon samples. An immunochemical approach is most commonly used for the analysis of glucagon due to the availability of an array of such methods for glucagon, including radioimmunoassay (RIA) and ELISAs. However, developments in technology for MS and chromatography, leading to an increased sensitivity and resolution, have opened greater opportunities for the application of LC–MS/MS methodology in peptide ana lysis. Moreover, glucagon is a good candidate for analysis by LC–MS/MS, for reasons that will be discussed.
10.4155/BIO.13.264 © 2013 Future Science Ltd
Bioanalysis (2013) 5(23), 2957–2972
Veniamin N Lapko*, Patrick S Miller, G Paul Brown, Rafiqul Islam, Sarah K Peters, Richard L Sukovaty, Peggy F Ruhn & Chris J Kafonek Celerion Inc., 621 Rose Street, Lincoln, NE 68502, USA *Author for correspondence: Tel.: +1 402 437 4867 Fax: +1 402 939 0428 E-mail: [email protected]
ISSN 1757-6180
2957
R eview |
Lapko, Miller, Brown et al. Glicentin Oxyntomodulin 1
33
61
Glucagon
GLP-1 69 72 78
Gut/brain
GLP-2 107 NH2
126
Major proglucagon fragment
158 160
Pancreas
1 5 7 10 15 NH2-His-Ser-Gln-Gly-Thr-Phe-Thr-Ser-Asp-Tyr-Ser-Lys-Tyr-Leu-Asp16 20 29 Ser-Arg-Arg-Ala-Gln-Asp-Phe-Val-Gln-Trp-Leu-Met-Asn-Thr-COOH
Figure 1. Proglucagon processing in the gut and pancreas, and amino acid sequence of glucagon.
Data on glucagon plasma concentrations (compiled in [8]) obtained predominantly by immunochemical methods vary significantly. Therefore, it is reasonable to suggest that immunochemical and MS methods for glucagon analysis could produce different results. Immunochemical methods rely on structural and functional integrity of the analyte and all of the antibodies, while MS methods require integrity of just the primary structure (i.e., chemical composition) of the analyte itself. Other issues, including but not limited to cross-reactivities of antibodies (in immunochemical methods) or poorly tracked, differential matrix effects (in LC–MS/MS methods), could also lead to disparate results. However, if an analytical (immunochemical or LC–MS/MS) method is working as intended, without hindrance from any methodological pitfalls, it is also reasonable to suggest that immunochemical and MS methods for the analysis of glucagon should produce consistent results to each other. Potential issues that may arise in the analysis of glucagon by immunochemical and MS-based methods are discussed, with suggestions for future directions in both methodological approaches. Considerations for glucagon analysis Whether an immunochemical or a LC–MS/MS approach is used for the analysis of glucagon, there are multiple factors to take into consideration for the development of an analytical method. The intrinsic physicochemical properties and characteristics of glucagon as a peptide are the basis for many of these considerations. 2958
Bioanalysis (2013) 5(23)
Because glucagon is an endogenous peptide in human plasma, consideration must be taken in the source of blank control matrix used for the analytical method. Depletion of glucagon in plasma can be performed using immunosorbents [9]. For LC–MS/MS methods, depletion of glucagon by incubation of plasma at room temperature for approximately 48 h is preferable to the addition of enzymes to the plasma for degradation of the peptide, as the former provides significantly lower background in extracted plasma [10]. However, there is no reason bovine serum albumin (BSA) solution could not be employed as a surrogate control matrix, as long as parallelism of calibration curves in BSA solution versus plasma is demonstrated. Reference
material Reference materials of different origin have been used for glucagon methods. These include purified porcine glucagon, recombinant glucagon and synthetic glucagon reference standards. The type of reference material is not as critical for LC–MS/MS methods as it is for immunochemical-based methods once a proper certification has been performed. In RIAs, the type of glucagon preparation used in an assay could affect, for instance, the quality of the tracer and the performance of the antibodies. Synthetic glucagon and recombinant glucagon reference materials from a variety of commercial sources were compared. European Pharmacopoeia recombinant glucagon reference material and recombinant glucagon from Eli Lilly’s glucagon emergency kit were found to correlate perfectly. However, some preparations of synthetic future science group
Sensitive glucagon quantification by immunochemical & LC–MS/MS methods glucagon, depending on the source and grade, may not correlate well to the European reference glucagon. The most common impurity found in synthetic glucagon preparations was DThr-5 glucagon, which was not always shown in a Certificate of Analysis. Up to 5.4% of DThr-5 glucagon was estimated to be present in some commercial preparations of synthetic glucagon [10]. Stock
solutions of glucagon Limits in solubility and the possibility for adsorption to containers should be considered before choosing an appropriate buffer or solvent composition for the preparation of stock solutions of glucagon. Glucagon is soluble below pH 3.5 and above pH 8.5. However, glucagon tends to aggregate in aqueous solutions [11]. Around neutral pH, detergents (e.g., Tween-20 or TritonX100) may be added to maintain solubility of glucagon. Typical buffers for immunoassays of glucagon are around pH 8.8 and include 0.5–1% protease-free BSA to prevent losses by adsorption. Just as glucagon in stocks made from pure glucagon reference materials are at risk of adsorption to unprotected containers, highly purified extracts (such as extracts by SPE for LC–MS/MS analysis) of glucagon from plasma or serum samples are also at risk. However, incidents involving significant adsorptive losses of glucagon during resolubilization of dried samples following partial purification of glucagon (such as by protein precipitation procedures) have not been reported. Most likely, this is because adsorption of glucagon is prevented by the presence of a significant amount of plasma components, serving as ‘keepers’ that adsorb in place of glucagon in the partially purified samples. Various approaches can be used to prepare glucagon stock and substock solutions, which are compatible with analysis by LC–MS/MS [12]. Neutral detergents may be added to samples of stock or substock solutions of glucagon, which are intended for analysis by LC–MS/MS, as long as a SPE is performed on the samples beforehand [10]. For stocks and substocks of glucagon that need to be injected onto a column without prior extraction, the peptide was determined to be stable in BSA-treated polypropylene tubes when dissolved in a solution of 25:75 acetonitrile (ACN):H2O containing 0.1% formic acid. For stock solutions of glucagon at concentrations below 100 µg/ml, the above solvent was supplemented with aprotinin (at 100 µg/ml) as a keeper peptide to ensure stability of glucagon even when future science group
| R eview
diluted to pg/ml substock concentrations [10]. An analogous approach was used for the preparation of insulin substocks and standards, for which 10 µg/ml of glargine (an insulin analogue) was added to the solutions [13]. Interestingly, glucagon itself at a concentration of 15 µg/ml has been used in solvents to minimize adsorption of other peptides during LC–MS/MS analysis [14]. The use of polar, aprotic solvents for stabilization of glucagon solutions was also recently reported [15]. Stability
of glucagon Glucagon has a half-life of 5–6 min in humans [16]. The liver and kidneys are primarily responsible for the degradation of glucagon in vivo [5,17]. Circulating glucagon is also degraded, largely through the proteolytic action of DPPIV [18]. Therefore, the serine protease inhibitor, aprotinin, at concentrations of 250–500 KIU, is commonly added to plasma samples for glucagon analysis. Reports on the stability of glucagon in human plasma are somewhat inconsistent. Semi-quantitative MALDI-TOF-MS data suggested that while the half-life of glucagon in EDTA plasma at ambient temperature is less than 6 h, if blood samples are collected in P800 vacutainers, which contain a proprietary cocktail including an inhibitor DPP-IV along with an esterase and other protease inhibitors, glucagon becomes significantly stabilized in the plasma, with the half-life increasing to over 48 h [19]. On the other hand, results from experiments by Sloan et al. implicate an insignificant role of the inhibitors used in the P800 cocktail [20]. While endogenous glucagon is relatively stable in human plasma, requiring incubation at room temperature for 48 h or more to completely deplete the plasma of endogenous glucagon, exogenous glucagon fortified into human plasma was found to be significantly less stable [Lapko V, Unpublished Data]. Supplementation of plasma samples with the water-soluble inhibitors leupeptin and 4-(2-aminoethyl) benzenesulfonyl fluoride – in addition to aprotinin – was found to significantly enhance the stability of glucagon fortified into human plasma. Plasma samples supplemented only with aprotinin exhibited a loss of over 65% of the fortified glucagon after 17 h of incubation in an ice water bath. However, over 97% of the glucagon remained intact in plasma samples containing a cocktail of aprotinin, leupeptin and 4-(2-aminoethyl) benzenesulfonyl fluoride [10]. Differences in the protein-binding state or structural conformation www.future-science.com
2959
R eview |
Lapko, Miller, Brown et al.
Key Term Electrochemiluminescence assay: Ligand-binding assay format in which the intensity of light produced by chemiluminescence is used for detection; chemiluminescence is stimulated by applying an electric potential to an appropriate label attached to a detection antibody.
2960
of endogenous glucagon versus exogenous gluca gon may explain the discrepancies in stability. The possibility for differences in the stability of endogenous versus exogenous glucagon in plasma samples should be carefully considered with regard to the evaluation of short-term and long-term stability of glucagon in QC samples fortified with exogenous glucagon. Bioanalysis of glucagon using ligand-binding assays The first glucagon immunoassay, described more than a half century ago by Unger et al. utilized rabbit antiserum against beef–pork glucagon [21]. Tracer 131I-glucagon was dialyzed and treated with an anion-exchange resin to remove nonpeptide radioactivity. Paper electrophoresis or sheep antirabbit serum were applied to separate the free and bound tracer. Sensitivity of 50 pg/ml glucagon was reported as the LOD for the assay. This work provided a valuable quantitative tool for investigation of the physiological role of glucagon. However, concerns regarding the reliability of the glucagon RIA arose due to noncontrolled interferences from plasma [22]. Matrix interferences, which are one of the primary concerns in an immunochemical assay, were addressed by Alford et al., who performed a comparative analysis that suggested a mean increase of 14 pg/ml plasma glucagon in nonketotic, non-obese diabetics over mean glucagon levels in healthy subjects (24 pg/ml) could be a biologically significant factor contributing to fasting hyperglycemia [9]. In those experiments, blank matrix was prepared from a portion of each unique plasma sample by depletion of endogenous glucagon with an immunosorbent. Separate calibration curves were prepared using each blank (glucagon-depleted) plasma sample for quantification of glucagon in the corresponding non-depleted plasma samples. However, the authors did not find a significant difference between samples quantified against calibration curves prepared with the immunosorbent-depleted plasma versus samples quantified against calibration curves prepared from pooled plasma depleted of glucagon via degradation over time [9]. Variations between different sources of antisera are another major concern in immunochemical assays. The use of different antisera, each directed against the C-terminal region of glucagon and each having no cross-reactivity to glucagon-like peptides in the gut, resulted in different concentrations of glucagon in patient Bioanalysis (2013) 5(23)
samples [23]. The difference in concentration was smaller when a preliminary acetone extraction was performed. With the development of improved immunization techniques and more available sources of antiglucagon antibodies, it became even more evident that similar immunoreactivity between different antiglucagon antibodies in plasma of normal subjects and patients with glucagon level disturbances did not guarantee analogous immunoreactivity in plasma samples from some hyperglucagonemic patients [24]. Many currently available RIA kits are able to quantify glucagon within a range of approximately 20–400 pg/ml, with selectivity against oxyntomodulin and insulin, as well as some other related peptides, dependent on the kit. Methods based on an ELISA offer a wider calibration range with a higher throughput of analysis than RIA-based methods. Despite the relatively small size of glucagon, a number of sandwich-type ELISA methods have been developed. An ELISA kit offered by R&D Systems can be used to measure glucagon concentrations from 31.3 to 2000 pg/ml in serum, plasma or cell culture samples without preliminary sample clean-up. The kit includes microplates coated with mouse monoclonal antiglucagon antibodies and mouse monoclonal antibodies conjugated to horseradish peroxidase for colorimetric detection. Plasma samples with glucagon concentrations between 315 and 1024 pg/ml were assayed with precision values better than 4% within an assay and better than 9% between assays [25]. A chemiluminescent kit by Millipore has a calibration range from 20.0 to 2000 pg/ml. The kit utilizes a specific antiglucagon capture antibody for binding to a microtiter plate coated with anchor antibodies, along with a biotinylated antiglucagon detection antibody for luminol detection [8,26]. Preliminary clean-up of plasma samples is required to ensure satisfactory performance of the kit. Although sample extraction prior to analysis could introduce additional variability into a method, a reproducible sample purification step may be beneficial for alleviation of matrix effect caused by the plasma. QC samples for the Millipore chemiluminescent assay are not plasma samples; instead, they are glucagon dissolved in the assay buffer. Similar to the calibration standards, the QC samples do not go through the extraction procedure. However, mean recoveries of 93.0 and 83.7% (with better than 3% CV) were demonstrated for plasma samples fortified with 44.0 and future science group
Sensitive glucagon quantification by immunochemical & LC–MS/MS methods
future science group
ELISA in Figure 2) versus a competitive RIA kit (x-axis) resulted in high correlation (R = 0.97) between the methods. However, the concentrations determined by the two methods differed by 2–3.8-fold, depending on the particular sample. Discrepancies between immunoassays are a rather common phenomenon for bioanalysis of peptides (even with the use of identical reference materials) and small molecules due to different cross-reactivity and other matrix interferences, which may differentially affect each particular immunoassay [30–33]. A summary of some of the methods featured above is highlighted in Table 1. Each of the above described methods can be placed into one of two broad categories: two-site sandwich assay and radioimmunoassay. There are limitations and advantages associated with each approach. A general comparison of these two approaches is summarized in Table 2 . Bioanalysis of glucagon by MS LC–MS/MS analysis of peptides Major issues in LC–MS/MS analysis of peptides are low sensitivity due to the formation of multiply charged ions and typically poor efficiency of SRM fragmentation, as well as adsorption of analytes to containers and instability of analytes in biological matrix. Therefore, development of highly sensitive LC–MS/MS-based methods often requires that careful attention be given to 250 y = 0.2986x + 7.9786
200
ELISA (pmol/l)
248 pg/ml glucagon, respectively. The precision for quantification of replicate plasma or serum samples with concentrations over 20 pg/ml was better than 5% CV [8,26]. Similarly, method performance in RIA kits is typically measured through QC samples prepared in an assay buffer [27,28]. Although QC samples prepared in the true biological matrix would generally be considered superior to QC samples prepared in an assay buffer for any bioanalytical method, incorporation of plasma QC samples into immunochemical assays may not be straightforward [29]. The sensitivity and specificity of available RIA and ELISA methods may still be insufficient for gaining a better understanding of the physiological roles of glucagon. Sensitivity and specificity issues were addressed by Sloan et al. in an immunochemical sandwich electrochemiluminescence (ECL) assay through the development of higher affinity monoclonal antiglucagon (mid-domain and C-terminus) antibodies, by way of saturation mutagenesis of the complementarity determining regions of heavy and light immunoglobulin chains [20]. The overall improvement of affinity to glucagon was over 100-fold for antibodies to the glucagon mid-domain and over 800-fold for antibodies to the C-terminal region of glucagon. Optimal pairing for the sandwich ECL assay was obtained when antiglucagon middomain antibody was used to capture glucagon and antiglucagon C-terminus antibody was used for detection. Even though the capture antibody showed cross-reactivity with oxyntomodulin, the detection antibody was highly specific for the C-terminus of glucagon. As a result, there was no cross-reactivity of the assay with oxyntomodulin, as well as some other incretins at concentrations of 100 ng/ml. Dilution of samples by eightfold with an assay buffer containing a heterophilic blocking reagent further improved method performance and reliability. Using the Meso Scale Discovery (MSD®) platform, an LLOQ of approximately 4 pg/ml was reported for the method, with a broad dynamic range of over 10,000-fold. Intra-batch precision of the assay for human plasma samples was 4.2, 3.7 and 2.4% for 104, 348 and 1040 pg/ml glucagon, respectively. Inter-batch precision for the same samples was from 15.1 to 9.7%. One drawback of this method is that it will measure the population of glucagon fragments that lack the N-terminal region. Comparative analysis of glucagon concentrations in samples measured by the ECL immunoassay (y-axis, labeled as
| R eview
R2 = 0.9467
150
100
50
0 0
200 400 600 Radioimmunoassay (pmol/l)
800
Figure 2. Correlation of glucagon concentrations in 26 samples determined by the sandwich electrochemiluminescence immunoassay and a competitive radioimmunoassay. Reprinted with permission from [20] © Elsevier, Canadian Society of Clinical Chemists (2012).
www.future-science.com
2961
R eview |
Lapko, Miller, Brown et al. each step involved in the method; that is, sample preparation, separation by LC, detection by MS and selection of the appropriate IS. The extraction procedure should have high recovery and strong reproducibility, independent of variations in the matrix, such as lipemic or hemolyzed plasma samples. Furthermore, the extraction procedure should complement the LC–MS conditions in terms of elimination or significant minimization of matrix effect and interferences. An appropriately chosen extraction procedure could provide an efficient extra dimension toward achieving superior selectivity as well as high sensitivity in the analysis. Protein precipitation is a straightforward procedure for sample preparation, which is routinely used in bioanalytical laboratories. When high concentrations of organic solvent are added to plasma samples, protein binding is typically minimal. However, losses of analyte by adsorption to precipitated proteins cannot necessarily be disregarded. SPE generally provides significantly greater purification of samples in comparison to protein precipitation, and SPE has been successfully applied in sample preparation of large peptides. Mixed-mode SPE sorbents are well-suited for the extraction of zwitterionic compounds, including peptides, as the sorbents utilize multiple modes of interaction with the analytes. The sorbents typically display low irreversible binding, even for extremely hydrophobic peptides, such as amyloid-b [34]. Neutral detergents can be added to samples to maintain peptide solubility, as they can be efficiently removed during subsequent washing steps with organic solvents [10]. Strong anion-exchange plates (in microelution
format) were used for the extraction of oxyntomodulin, with approximately 60% recovery. Acidified 75% ACN was employed for elution of the peptide after the sorbent was washed with a 5% ACN solution [35]. Weak cation exchange was successfully applied to the purification of large peptides up to 7800 Da [36]. An efficient organic solvent wash, while preserving ionic interaction of the peptide to the sorbent, was incorporated by Lövgren et al. in the development of a sensitive LC–MS/MS method for the peptide FE202158 [37]. Affinity chromatography, combined with LC– MS/MS, could provide superior selectivity, sensitivity and accuracy of analysis. Coupling affinity extraction with LC–MS/MS detection has been demonstrated for small molecules [38], for proteomic research [39], as well as for quantification of large peptides [40]. For the affinity capture of the insulinotropic peptides GLP-1 and GIP (and their incretin metabolites), Wolf et al. described the application of magnetic beads [41]. Anti-IgG antibodies were covalently bound to the magnetic beads, rendering the beads capable of binding to anti-GLP-1 and anti-GIP polyclonal antibodies with high affinity for the incretins. An LLOQ of approximately 25 pg/ml GIP, sufficient for analysis of basal levels, was demonstrated for the method using an API 150EX™ mass spectrometer [41]. More recently, an LC–MS/MS method [42] employing an affinity extraction on commercial mouse monoclonal antibody, covalently bound to Sepharose, was used in a pilot study for comparison of insulin immunoassays to an LC–MS/MS isotope-dilution method [13]. The number of options for the chromatographic approaches that may be used for
Table 1. Sensitivity and specificity of glucagon immunoassays. Range (pg/ml†) Assay type 25.0– 400
16.4–522 31.3–2000
20.0–2000
0.49–6790
Reported specificity/crossreactivity
RIA
Oxyntomodulin
Sensitive glucagon quantification by immunochemical and LC–MS/MS methods The peptide hormone glucagon plays an important role in homeostasis of glucose concentrations in the blood. Its biological importance is evidenced through the conservation of its peptide sequence between species. Reliable assays for glucagon in biological samples are important for gaining a better understanding of the pathology and treatment of diabetes. Numerous assays are available for the analysis of glucagon in biological samples, the majority of which employ an immunochemical approach and have been available for many years. However, recent advances in MS instrumentation and the amenability of glucagon for analysis by LC–MS/MS has brought these new methods to the forefront. Concentrations of glucagon determined from different methods are not always consistent and this review provides suggestions of how to improve the reliability of methods for glucagon analysis.
Diabetes is a group of metabolic diseases characterized by a high level of glucose in the blood. A significant research effort is currently directed towards understanding and treating diabetes, which was estimated to cost the USA approximately US$175 billion for direct and indirect healthcare in 2007 [1]. In 2012, nearly 10,000 articles with diabetes in the title (PubMed search) were published. Insulin and glucagon are peptide hormones that are responsible for the maintenance of normal glucose levels in blood. Insulin acts in response to an elevated blood glucose concentration by increasing the uptake of glucose by cells, while the primary role of glucagon is to respond to hypoglycemia by increasing glycogenolysis and gluconeogenesis in the liver in order to increase the blood glucose concentration [2]. Elevated glucagon levels are suggested to contribute to the severity of hyperglycemia in diabetic patients [3]. As a therapeutic agent, glucagon may be administered as an emergency treatment to combat severe hypoglycemia associated with the use of insulin and some other medications for diabetic patients. Glucagon (a peptide chain of 29 amino acids) and two other peptide hormones, GLP-1 and GLP-2, are encoded by a single proglucagon gene in humans. Tissue-specific, post-translational processing of the proglucagon gene product, from the transcript, results in the secretion of glucagon by pancreatic a-cells, whereas intestinal b-cells produce GLP-1 and GLP-2, as reviewed by Pocai [4]. Proglucagon processing in the intestine also results in two other peptides (enteroglucagon) that contain the
glucagon sequence – oxyntomodulin and glicentin (Figure 1). The peptide sequence of glucagon is highly conserved between mammalian species, which reinforces the notion of its important biological functions. Glucagon receptors are not only expressed in the liver, but also in the kidneys, pancreas, intestines, brain and adipose tissues, as reviewed by Authier and Desbuquois [5]. Experimental data suggest that glucagon performs many other biological functions, in addition to acting as an antihypoglycemic agent, and plays an important role in stress response and energy metabolism [6,7]. Changes in glucagon concentrations as a result of dietary, metabolic and pharmaceutical interventions can provide insight into the epidemiology of diabetes and the efficacy of potential treatments. Unraveling the multifaceted functionality of glucagon strongly relies on the availability of accurate, precise and robust bioa nalytical methods. In addition, sensitivity at a low pg/ml level is required to analyze endogenous glucagon samples. An immunochemical approach is most commonly used for the analysis of glucagon due to the availability of an array of such methods for glucagon, including radioimmunoassay (RIA) and ELISAs. However, developments in technology for MS and chromatography, leading to an increased sensitivity and resolution, have opened greater opportunities for the application of LC–MS/MS methodology in peptide ana lysis. Moreover, glucagon is a good candidate for analysis by LC–MS/MS, for reasons that will be discussed.
10.4155/BIO.13.264 © 2013 Future Science Ltd
Bioanalysis (2013) 5(23), 2957–2972
Veniamin N Lapko*, Patrick S Miller, G Paul Brown, Rafiqul Islam, Sarah K Peters, Richard L Sukovaty, Peggy F Ruhn & Chris J Kafonek Celerion Inc., 621 Rose Street, Lincoln, NE 68502, USA *Author for correspondence: Tel.: +1 402 437 4867 Fax: +1 402 939 0428 E-mail: [email protected]
ISSN 1757-6180
2957
R eview |
Lapko, Miller, Brown et al. Glicentin Oxyntomodulin 1
33
61
Glucagon
GLP-1 69 72 78
Gut/brain
GLP-2 107 NH2
126
Major proglucagon fragment
158 160
Pancreas
1 5 7 10 15 NH2-His-Ser-Gln-Gly-Thr-Phe-Thr-Ser-Asp-Tyr-Ser-Lys-Tyr-Leu-Asp16 20 29 Ser-Arg-Arg-Ala-Gln-Asp-Phe-Val-Gln-Trp-Leu-Met-Asn-Thr-COOH
Figure 1. Proglucagon processing in the gut and pancreas, and amino acid sequence of glucagon.
Data on glucagon plasma concentrations (compiled in [8]) obtained predominantly by immunochemical methods vary significantly. Therefore, it is reasonable to suggest that immunochemical and MS methods for glucagon analysis could produce different results. Immunochemical methods rely on structural and functional integrity of the analyte and all of the antibodies, while MS methods require integrity of just the primary structure (i.e., chemical composition) of the analyte itself. Other issues, including but not limited to cross-reactivities of antibodies (in immunochemical methods) or poorly tracked, differential matrix effects (in LC–MS/MS methods), could also lead to disparate results. However, if an analytical (immunochemical or LC–MS/MS) method is working as intended, without hindrance from any methodological pitfalls, it is also reasonable to suggest that immunochemical and MS methods for the analysis of glucagon should produce consistent results to each other. Potential issues that may arise in the analysis of glucagon by immunochemical and MS-based methods are discussed, with suggestions for future directions in both methodological approaches. Considerations for glucagon analysis Whether an immunochemical or a LC–MS/MS approach is used for the analysis of glucagon, there are multiple factors to take into consideration for the development of an analytical method. The intrinsic physicochemical properties and characteristics of glucagon as a peptide are the basis for many of these considerations. 2958
Bioanalysis (2013) 5(23)
Because glucagon is an endogenous peptide in human plasma, consideration must be taken in the source of blank control matrix used for the analytical method. Depletion of glucagon in plasma can be performed using immunosorbents [9]. For LC–MS/MS methods, depletion of glucagon by incubation of plasma at room temperature for approximately 48 h is preferable to the addition of enzymes to the plasma for degradation of the peptide, as the former provides significantly lower background in extracted plasma [10]. However, there is no reason bovine serum albumin (BSA) solution could not be employed as a surrogate control matrix, as long as parallelism of calibration curves in BSA solution versus plasma is demonstrated. Reference
material Reference materials of different origin have been used for glucagon methods. These include purified porcine glucagon, recombinant glucagon and synthetic glucagon reference standards. The type of reference material is not as critical for LC–MS/MS methods as it is for immunochemical-based methods once a proper certification has been performed. In RIAs, the type of glucagon preparation used in an assay could affect, for instance, the quality of the tracer and the performance of the antibodies. Synthetic glucagon and recombinant glucagon reference materials from a variety of commercial sources were compared. European Pharmacopoeia recombinant glucagon reference material and recombinant glucagon from Eli Lilly’s glucagon emergency kit were found to correlate perfectly. However, some preparations of synthetic future science group
Sensitive glucagon quantification by immunochemical & LC–MS/MS methods glucagon, depending on the source and grade, may not correlate well to the European reference glucagon. The most common impurity found in synthetic glucagon preparations was DThr-5 glucagon, which was not always shown in a Certificate of Analysis. Up to 5.4% of DThr-5 glucagon was estimated to be present in some commercial preparations of synthetic glucagon [10]. Stock
solutions of glucagon Limits in solubility and the possibility for adsorption to containers should be considered before choosing an appropriate buffer or solvent composition for the preparation of stock solutions of glucagon. Glucagon is soluble below pH 3.5 and above pH 8.5. However, glucagon tends to aggregate in aqueous solutions [11]. Around neutral pH, detergents (e.g., Tween-20 or TritonX100) may be added to maintain solubility of glucagon. Typical buffers for immunoassays of glucagon are around pH 8.8 and include 0.5–1% protease-free BSA to prevent losses by adsorption. Just as glucagon in stocks made from pure glucagon reference materials are at risk of adsorption to unprotected containers, highly purified extracts (such as extracts by SPE for LC–MS/MS analysis) of glucagon from plasma or serum samples are also at risk. However, incidents involving significant adsorptive losses of glucagon during resolubilization of dried samples following partial purification of glucagon (such as by protein precipitation procedures) have not been reported. Most likely, this is because adsorption of glucagon is prevented by the presence of a significant amount of plasma components, serving as ‘keepers’ that adsorb in place of glucagon in the partially purified samples. Various approaches can be used to prepare glucagon stock and substock solutions, which are compatible with analysis by LC–MS/MS [12]. Neutral detergents may be added to samples of stock or substock solutions of glucagon, which are intended for analysis by LC–MS/MS, as long as a SPE is performed on the samples beforehand [10]. For stocks and substocks of glucagon that need to be injected onto a column without prior extraction, the peptide was determined to be stable in BSA-treated polypropylene tubes when dissolved in a solution of 25:75 acetonitrile (ACN):H2O containing 0.1% formic acid. For stock solutions of glucagon at concentrations below 100 µg/ml, the above solvent was supplemented with aprotinin (at 100 µg/ml) as a keeper peptide to ensure stability of glucagon even when future science group
| R eview
diluted to pg/ml substock concentrations [10]. An analogous approach was used for the preparation of insulin substocks and standards, for which 10 µg/ml of glargine (an insulin analogue) was added to the solutions [13]. Interestingly, glucagon itself at a concentration of 15 µg/ml has been used in solvents to minimize adsorption of other peptides during LC–MS/MS analysis [14]. The use of polar, aprotic solvents for stabilization of glucagon solutions was also recently reported [15]. Stability
of glucagon Glucagon has a half-life of 5–6 min in humans [16]. The liver and kidneys are primarily responsible for the degradation of glucagon in vivo [5,17]. Circulating glucagon is also degraded, largely through the proteolytic action of DPPIV [18]. Therefore, the serine protease inhibitor, aprotinin, at concentrations of 250–500 KIU, is commonly added to plasma samples for glucagon analysis. Reports on the stability of glucagon in human plasma are somewhat inconsistent. Semi-quantitative MALDI-TOF-MS data suggested that while the half-life of glucagon in EDTA plasma at ambient temperature is less than 6 h, if blood samples are collected in P800 vacutainers, which contain a proprietary cocktail including an inhibitor DPP-IV along with an esterase and other protease inhibitors, glucagon becomes significantly stabilized in the plasma, with the half-life increasing to over 48 h [19]. On the other hand, results from experiments by Sloan et al. implicate an insignificant role of the inhibitors used in the P800 cocktail [20]. While endogenous glucagon is relatively stable in human plasma, requiring incubation at room temperature for 48 h or more to completely deplete the plasma of endogenous glucagon, exogenous glucagon fortified into human plasma was found to be significantly less stable [Lapko V, Unpublished Data]. Supplementation of plasma samples with the water-soluble inhibitors leupeptin and 4-(2-aminoethyl) benzenesulfonyl fluoride – in addition to aprotinin – was found to significantly enhance the stability of glucagon fortified into human plasma. Plasma samples supplemented only with aprotinin exhibited a loss of over 65% of the fortified glucagon after 17 h of incubation in an ice water bath. However, over 97% of the glucagon remained intact in plasma samples containing a cocktail of aprotinin, leupeptin and 4-(2-aminoethyl) benzenesulfonyl fluoride [10]. Differences in the protein-binding state or structural conformation www.future-science.com
2959
R eview |
Lapko, Miller, Brown et al.
Key Term Electrochemiluminescence assay: Ligand-binding assay format in which the intensity of light produced by chemiluminescence is used for detection; chemiluminescence is stimulated by applying an electric potential to an appropriate label attached to a detection antibody.
2960
of endogenous glucagon versus exogenous gluca gon may explain the discrepancies in stability. The possibility for differences in the stability of endogenous versus exogenous glucagon in plasma samples should be carefully considered with regard to the evaluation of short-term and long-term stability of glucagon in QC samples fortified with exogenous glucagon. Bioanalysis of glucagon using ligand-binding assays The first glucagon immunoassay, described more than a half century ago by Unger et al. utilized rabbit antiserum against beef–pork glucagon [21]. Tracer 131I-glucagon was dialyzed and treated with an anion-exchange resin to remove nonpeptide radioactivity. Paper electrophoresis or sheep antirabbit serum were applied to separate the free and bound tracer. Sensitivity of 50 pg/ml glucagon was reported as the LOD for the assay. This work provided a valuable quantitative tool for investigation of the physiological role of glucagon. However, concerns regarding the reliability of the glucagon RIA arose due to noncontrolled interferences from plasma [22]. Matrix interferences, which are one of the primary concerns in an immunochemical assay, were addressed by Alford et al., who performed a comparative analysis that suggested a mean increase of 14 pg/ml plasma glucagon in nonketotic, non-obese diabetics over mean glucagon levels in healthy subjects (24 pg/ml) could be a biologically significant factor contributing to fasting hyperglycemia [9]. In those experiments, blank matrix was prepared from a portion of each unique plasma sample by depletion of endogenous glucagon with an immunosorbent. Separate calibration curves were prepared using each blank (glucagon-depleted) plasma sample for quantification of glucagon in the corresponding non-depleted plasma samples. However, the authors did not find a significant difference between samples quantified against calibration curves prepared with the immunosorbent-depleted plasma versus samples quantified against calibration curves prepared from pooled plasma depleted of glucagon via degradation over time [9]. Variations between different sources of antisera are another major concern in immunochemical assays. The use of different antisera, each directed against the C-terminal region of glucagon and each having no cross-reactivity to glucagon-like peptides in the gut, resulted in different concentrations of glucagon in patient Bioanalysis (2013) 5(23)
samples [23]. The difference in concentration was smaller when a preliminary acetone extraction was performed. With the development of improved immunization techniques and more available sources of antiglucagon antibodies, it became even more evident that similar immunoreactivity between different antiglucagon antibodies in plasma of normal subjects and patients with glucagon level disturbances did not guarantee analogous immunoreactivity in plasma samples from some hyperglucagonemic patients [24]. Many currently available RIA kits are able to quantify glucagon within a range of approximately 20–400 pg/ml, with selectivity against oxyntomodulin and insulin, as well as some other related peptides, dependent on the kit. Methods based on an ELISA offer a wider calibration range with a higher throughput of analysis than RIA-based methods. Despite the relatively small size of glucagon, a number of sandwich-type ELISA methods have been developed. An ELISA kit offered by R&D Systems can be used to measure glucagon concentrations from 31.3 to 2000 pg/ml in serum, plasma or cell culture samples without preliminary sample clean-up. The kit includes microplates coated with mouse monoclonal antiglucagon antibodies and mouse monoclonal antibodies conjugated to horseradish peroxidase for colorimetric detection. Plasma samples with glucagon concentrations between 315 and 1024 pg/ml were assayed with precision values better than 4% within an assay and better than 9% between assays [25]. A chemiluminescent kit by Millipore has a calibration range from 20.0 to 2000 pg/ml. The kit utilizes a specific antiglucagon capture antibody for binding to a microtiter plate coated with anchor antibodies, along with a biotinylated antiglucagon detection antibody for luminol detection [8,26]. Preliminary clean-up of plasma samples is required to ensure satisfactory performance of the kit. Although sample extraction prior to analysis could introduce additional variability into a method, a reproducible sample purification step may be beneficial for alleviation of matrix effect caused by the plasma. QC samples for the Millipore chemiluminescent assay are not plasma samples; instead, they are glucagon dissolved in the assay buffer. Similar to the calibration standards, the QC samples do not go through the extraction procedure. However, mean recoveries of 93.0 and 83.7% (with better than 3% CV) were demonstrated for plasma samples fortified with 44.0 and future science group
Sensitive glucagon quantification by immunochemical & LC–MS/MS methods
future science group
ELISA in Figure 2) versus a competitive RIA kit (x-axis) resulted in high correlation (R = 0.97) between the methods. However, the concentrations determined by the two methods differed by 2–3.8-fold, depending on the particular sample. Discrepancies between immunoassays are a rather common phenomenon for bioanalysis of peptides (even with the use of identical reference materials) and small molecules due to different cross-reactivity and other matrix interferences, which may differentially affect each particular immunoassay [30–33]. A summary of some of the methods featured above is highlighted in Table 1. Each of the above described methods can be placed into one of two broad categories: two-site sandwich assay and radioimmunoassay. There are limitations and advantages associated with each approach. A general comparison of these two approaches is summarized in Table 2 . Bioanalysis of glucagon by MS LC–MS/MS analysis of peptides Major issues in LC–MS/MS analysis of peptides are low sensitivity due to the formation of multiply charged ions and typically poor efficiency of SRM fragmentation, as well as adsorption of analytes to containers and instability of analytes in biological matrix. Therefore, development of highly sensitive LC–MS/MS-based methods often requires that careful attention be given to 250 y = 0.2986x + 7.9786
200
ELISA (pmol/l)
248 pg/ml glucagon, respectively. The precision for quantification of replicate plasma or serum samples with concentrations over 20 pg/ml was better than 5% CV [8,26]. Similarly, method performance in RIA kits is typically measured through QC samples prepared in an assay buffer [27,28]. Although QC samples prepared in the true biological matrix would generally be considered superior to QC samples prepared in an assay buffer for any bioanalytical method, incorporation of plasma QC samples into immunochemical assays may not be straightforward [29]. The sensitivity and specificity of available RIA and ELISA methods may still be insufficient for gaining a better understanding of the physiological roles of glucagon. Sensitivity and specificity issues were addressed by Sloan et al. in an immunochemical sandwich electrochemiluminescence (ECL) assay through the development of higher affinity monoclonal antiglucagon (mid-domain and C-terminus) antibodies, by way of saturation mutagenesis of the complementarity determining regions of heavy and light immunoglobulin chains [20]. The overall improvement of affinity to glucagon was over 100-fold for antibodies to the glucagon mid-domain and over 800-fold for antibodies to the C-terminal region of glucagon. Optimal pairing for the sandwich ECL assay was obtained when antiglucagon middomain antibody was used to capture glucagon and antiglucagon C-terminus antibody was used for detection. Even though the capture antibody showed cross-reactivity with oxyntomodulin, the detection antibody was highly specific for the C-terminus of glucagon. As a result, there was no cross-reactivity of the assay with oxyntomodulin, as well as some other incretins at concentrations of 100 ng/ml. Dilution of samples by eightfold with an assay buffer containing a heterophilic blocking reagent further improved method performance and reliability. Using the Meso Scale Discovery (MSD®) platform, an LLOQ of approximately 4 pg/ml was reported for the method, with a broad dynamic range of over 10,000-fold. Intra-batch precision of the assay for human plasma samples was 4.2, 3.7 and 2.4% for 104, 348 and 1040 pg/ml glucagon, respectively. Inter-batch precision for the same samples was from 15.1 to 9.7%. One drawback of this method is that it will measure the population of glucagon fragments that lack the N-terminal region. Comparative analysis of glucagon concentrations in samples measured by the ECL immunoassay (y-axis, labeled as
| R eview
R2 = 0.9467
150
100
50
0 0
200 400 600 Radioimmunoassay (pmol/l)
800
Figure 2. Correlation of glucagon concentrations in 26 samples determined by the sandwich electrochemiluminescence immunoassay and a competitive radioimmunoassay. Reprinted with permission from [20] © Elsevier, Canadian Society of Clinical Chemists (2012).
www.future-science.com
2961
R eview |
Lapko, Miller, Brown et al. each step involved in the method; that is, sample preparation, separation by LC, detection by MS and selection of the appropriate IS. The extraction procedure should have high recovery and strong reproducibility, independent of variations in the matrix, such as lipemic or hemolyzed plasma samples. Furthermore, the extraction procedure should complement the LC–MS conditions in terms of elimination or significant minimization of matrix effect and interferences. An appropriately chosen extraction procedure could provide an efficient extra dimension toward achieving superior selectivity as well as high sensitivity in the analysis. Protein precipitation is a straightforward procedure for sample preparation, which is routinely used in bioanalytical laboratories. When high concentrations of organic solvent are added to plasma samples, protein binding is typically minimal. However, losses of analyte by adsorption to precipitated proteins cannot necessarily be disregarded. SPE generally provides significantly greater purification of samples in comparison to protein precipitation, and SPE has been successfully applied in sample preparation of large peptides. Mixed-mode SPE sorbents are well-suited for the extraction of zwitterionic compounds, including peptides, as the sorbents utilize multiple modes of interaction with the analytes. The sorbents typically display low irreversible binding, even for extremely hydrophobic peptides, such as amyloid-b [34]. Neutral detergents can be added to samples to maintain peptide solubility, as they can be efficiently removed during subsequent washing steps with organic solvents [10]. Strong anion-exchange plates (in microelution
format) were used for the extraction of oxyntomodulin, with approximately 60% recovery. Acidified 75% ACN was employed for elution of the peptide after the sorbent was washed with a 5% ACN solution [35]. Weak cation exchange was successfully applied to the purification of large peptides up to 7800 Da [36]. An efficient organic solvent wash, while preserving ionic interaction of the peptide to the sorbent, was incorporated by Lövgren et al. in the development of a sensitive LC–MS/MS method for the peptide FE202158 [37]. Affinity chromatography, combined with LC– MS/MS, could provide superior selectivity, sensitivity and accuracy of analysis. Coupling affinity extraction with LC–MS/MS detection has been demonstrated for small molecules [38], for proteomic research [39], as well as for quantification of large peptides [40]. For the affinity capture of the insulinotropic peptides GLP-1 and GIP (and their incretin metabolites), Wolf et al. described the application of magnetic beads [41]. Anti-IgG antibodies were covalently bound to the magnetic beads, rendering the beads capable of binding to anti-GLP-1 and anti-GIP polyclonal antibodies with high affinity for the incretins. An LLOQ of approximately 25 pg/ml GIP, sufficient for analysis of basal levels, was demonstrated for the method using an API 150EX™ mass spectrometer [41]. More recently, an LC–MS/MS method [42] employing an affinity extraction on commercial mouse monoclonal antibody, covalently bound to Sepharose, was used in a pilot study for comparison of insulin immunoassays to an LC–MS/MS isotope-dilution method [13]. The number of options for the chromatographic approaches that may be used for
Table 1. Sensitivity and specificity of glucagon immunoassays. Range (pg/ml†) Assay type 25.0– 400
16.4–522 31.3–2000
20.0–2000
0.49–6790
Reported specificity/crossreactivity
RIA
Oxyntomodulin
