Article pubs.acs.org/ac
High-Throughput Differentiation of Heparin from Other Glycosaminoglycans by Pyrolysis Mass Spectrometry Peter Nemes,*,† William J. Hoover,† and David A. Keire‡ †
Division of Chemistry and Materials Science, Center for Devices and Radiological Health and ‡Division of Pharmaceutical Analysis, Center for Drug Evaluation and Research, Food and Drug Administration (FDA), United States S Supporting Information *
ABSTRACT: Sensors with high chemical specificity and enhanced sample throughput are vital to screening food products and medical devices for chemical or biochemical contaminants that may pose a threat to public health. For example, the rapid detection of oversulfated chondroitin sulfate (OSCS) in heparin could prevent reoccurrence of heparin adulteration that caused hundreds of severe adverse events including deaths worldwide in 2007−2008. Here, rapid pyrolysis is integrated with direct analysis in real time (DART) mass spectrometry to rapidly screen major glycosaminoglycans, including heparin, chondroitin sulfate A, dermatan sulfate, and OSCS. The results demonstrate that, compared to traditional liquid chromatography-based analyses, pyrolysis mass spectrometry achieved at least 250-fold higher sample throughput and was compatible with samples volume-limited to about 300 nL. Pyrolysis yielded an abundance of fragment ions (e.g., 150 different m/z species), many of which were specific to the parent compound. Using multivariate and statistical data analysis models, these data enabled facile differentiation of the glycosaminoglycans with high throughput. After method development was completed, authentically contaminated samples obtained during the heparin crisis by the FDA were analyzed in a blinded manner for OSCS contamination. The lower limit of differentiation and detection were 0.1% (w/w) OSCS in heparin and 100 ng/μL (20 ng) OSCS in water, respectively. For quantitative purposes the linear dynamic range spanned approximately 3 orders of magnitude. Moreover, this chemical readout was successfully employed to find clues in the manufacturing history of the heparin samples that can be used for surveillance purposes. The presented technology and data analysis protocols are anticipated to be readily adaptable to other chemical and biochemical agents and volume-limited samples.
B
0.1% by polyacrylamide gel electrophoresis (PAGE),15 0.5% by potentiometric polyanion-specific sensors,19 and ∼1% by near-infrared reflectance and Raman spectroscopy20 as well as agarose gel electrophoresis.21 In addition, reversed-phase ion pairing HPLC-MS analysis of disaccharide-level digests of heparin has been applied for detection of OSCS in heparin.8,22 Required sample amounts ranged from tens of milligrams by NMR to a few hundreds of micrograms by MS, demonstrating complementary performance characteristics for these protocols.23 MS proved particularly powerful because this approach separated the drug from the contaminants by HPLC and provided high-specificity detection and structural elucidation. Workflows by HPLC−MS required GAG chains, usually 10−50 kDa in size, to be digested to smaller disaccharide units (via enzymatic digestion) and then separated prior to detection by ion-pairing chromatography, amounting to typical analysis times from hours to days.5 While traditional analytical technologies and protocols underlie the modern-day quality-assurance of heparin and heparin-containing products, they require expert users and attain a sample throughput that is relatively low.
etween late 2007 and early 2008, adulterated heparin was associated with 94 deaths and 574 adverse events in the US with further adverse reactions registered worldwide upon administration of this drug.1−3 An international research effort by regulatory agencies, academia, and industry identified an oversulfated glycosaminoglycan (GAG) termed oversulfated chondroitin sulfate (OSCS) (Figure 1) as the major contaminant in the adulterated heparin.1,4 In response to the health crisis, the US and European pharmacopeias revised the heparin sodium monograph to require characterization of this material by analytical techniques that were capable of addressing the heterogeneous structures of GAGs (Figure 1), including nuclear magnetic resonance spectroscopy (NMR) and strong anion exchange high performance liquid chromatography (SAX-HPLC) coupled to ultraviolet spectroscopy and mass spectrometry (MS) detection.5−12 In addition to NMR and SAX-HPLC, many other orthogonal methods were developed for the detection of OSCS in heparin with different limits of detection (LOD). As even a trace amount of OSCS in heparin is evidence for a contamination of the heparin supply chain, the most sensitive method that can assess heparin quality on the most samples are desirable assay properties. LOD for OSCS in heparin (w/w) was 0.03% by SAX13 and weak anion exchange14 HPLC-UV, ∼0.1% by multidimensional NMR,15,16 0.05−0.1% by capillary electrophoresis (CE),17,18 This article not subject to U.S. Copyright. Published 2013 by the American Chemical Society
Received: May 1, 2013 Accepted: July 9, 2013 Published: July 9, 2013 7405
dx.doi.org/10.1021/ac401318q | Anal. Chem. 2013, 85, 7405−7412
Analytical Chemistry
Article
Figure 1. Mass spectral fingerprinting glycosaminoglycans in high-throughput by pyrolysis DART. Leading mechanisms of anion generation are interpreted for (left panel) heparin, dermatan sulfate, and the major adulterant of the heparin crisis, oversulfated chondroitin sulfate (OSCS). (Right panel) Major disaccharide units are shown for each GAG type (n denotes repeating units). Horizontal arrows indicate dehydration and tilted arrows designate oxidation. Detected ring scissions are marked according to the MS nomenclature of sugars. Key: R1, R2 = H or SO3H; Q = H, acetyl, or SO3H.
for GAG analysis but has yet to be made compatible with simple sample preparation, high-throughput sampling, and automated data analysis to enable screenings within an hour, preferably minutes. Ambient ionization MS presents an opportunity to increase overall screening throughput, albeit with intrinsic limitations specific to GAGs. These techniques usually require minimal or no sample preparation and allow for direct sampling and ionization, which in turn translate into fast measurements, typically with ∼1 min/sample throughput on small molecules.36−40 Of the more than thirty different types of ambient ion sources reported in recent years,36,39,40 desorption electrospray ionization41−43 and direct analysis in real time (DART)44−46 emerged as particularly powerful alternatives to traditional analytics of screenings.47 These techniques have been commercialized and shown to have high measurement capacity and an extensive track record in biomolecular and quality-type analyses. DART utilizes heated metastable gas (e.g., helium) to desorb and ionize compounds with molecular weights below ∼1 kDa via Penning ionization. Thus, the average molecular weights of heparin API and its OSCS contaminants (∼18 kDa) are outside the ordinary mass range accessible by DART MS. Here we integrate rapid pyrolysis with DART MS and multivariate data analysis to extend high-throughput measurements to large GAGs including the authentic adulterated heparin from the 2007−2008 health crisis. Our experiments capitalize on a commercial DART ion source that is tuned to thermally degrade large polysaccharide chains of major GAGs, specifically heparin, chondroitin sulfate A (CSA), dermatan sulfate (DS), and OSCS. The approach achieves at least 250fold higher measurement throughput than feasible by current protocols of heparin screening using LC-MS and sourceinduced dissociation MS. Furthermore, it consumes less than 300 nL of solution, facilitating analyses on volume-limited or expensive samples. The pyrolysis products are structurally representative of the parent compound and, when combined
With ubiquitous usage and internationally dispersed manufacturing sites, quality assurance of the heparin supply chain would be further improved with increasing measurement throughput and detection sensitivity and specificity. This is true not only for formulated heparin drug products but also for the more than 200 medical devices and diagnostic products that contain or may be coated with heparin.24 Since the early characterization of its anticoagulant and antithrombotic properties on mammalian models as well as use in humans in the 1930s,25 heparin consumption has amounted to tons of the active pharmaceutical ingredient (API) each year, making it critical to improve the available tests to ensure the quality of this heavily used drug. Currently, there is a high and an insofar unmet demand for measurements capable of fast, sensitive, and chemically specific GAG detection while requiring minimal user expertise. Recent advances in technology began affording moderate screening throughput for GAGs by simplifying sample preparation and detection as well as automating data analysis. By combining spectral libraries with multivariate data analysis, NMR rapidly differentiated contaminated heparin samples.26,27 Potentiometric sensors28 and calorimetric assays employing supramolecular interactions with enzymes29−32 and self-assembling ligands33 provided fast operation and good LODs using instruments that are commonly present in laboratories. Of particular note was an assay based on a cationic polythiophene polymer that changed color upon association with anions in a GAG-specific manner34 and enabled measuring OSCS in heparin with a LOD of 0.1% (w/w) by the unaided eye and 0.003% using a plate reader.23 Likewise, GAG measurements by MS were recently simplified using source-induced dissociation with a LOD of 0.5% chondroitin/dermatan sulfate and 5% heparan sulfate in heparin and a potential to detect OSCS.35 To simplify ion generation and manual data interpretation, samples were thoroughly desalted in this approach, improving the net screening throughput to few hours per sample. Chemically selective detection by MS offers unique advantages 7406
dx.doi.org/10.1021/ac401318q | Anal. Chem. 2013, 85, 7405−7412
Analytical Chemistry
Article
with multivariate and statistical analysis models, successfully differentiate the GAGs including heparin standards from the adulterated heparin. Moreover, this rich chemical information can be used as a fingerprint to evaluate the manufacturing history of samples. Thus, pyrolysis DART MS and the presented data analysis models are a viable alternative to established and emerging analytics for regulatory screenings on GAGs.
Standard safety protocols were followed when working with chemicals.
■
RESULTS AND DISCUSSION This work presents simplified sample preparation and rapid, ambient pyrolisis as a means to extend GAG screenings to highthroughput DART MS detection. Our workflow of sample preparation intentionally excluded time-consuming steps that would be normally required in mass spectrometric analysis of GAGs, such as desalting, enzymatic/chemical digestion, and purification in LC-MS and/or in-source fragmentation MS. Powdered sodium salts of GAGs were dissolved in aqueous solvents (e.g., water) without further modification. A systematic study revealed that less than ∼300 nL of these solutions could be transferred and rapidly introduced into the DART ion source, typically within 3 s, by dipping a glass capillary in the sample within 1 s (Figure S1, SI). Metastable helium heated to 100−500 °C was generated using the DART source and employed to pyrolize GAGs from the capillaries. MS analysis of the generated ions showed that some GAG monosaccharides pyrolized at lower temperatures (e.g., 250 °C for N-acetylgalactosamine (GalNac)), whereas other GAGs required harsher conditions (e.g., 450 °C for OSCS). Pyrolysis was completed within ∼10 s, yielding a net measurement throughput of approximately 30 s/sample (Figure S2). Technical specifications regarding capillary-based sampling and pyrolysis are discussed in the SI. Compared to GAG analysis by LC−MS and source-induced dissociation MS, which can take from 2 h to a day and require tens of μL of sample, pyrolysis DART MS improved the net measurement throughput by more than 250fold and enabled compatibility with volume-limited samples in the sub-μL regime. Each type of GAG generated unique pyrolysis products upon MS detection, which were dependent on their structures (Figure 1). To help to deduce mechanisms of ion generation via pyrolysis DART, GAG monosaccharide subunits were evaluated against soft-ionization ESI. The mass spectral fingerprints of glucuronic acid (GlcA) and N-acetylgalactosamine-6-sulfate (GalNAc6S) are shown in Figure S3. Single- and multistage mass measurements (MS2−4) and isotope distribution analysis of ion signals revealed that pyrolysis occurred via a combination of parallel and sequential reactions. Common pathways included losses of neutral molecules such as water, hydrogen, sulfur trioxide, and ketene as well as charged species ranging from the sulfur trioxide and sulfur dioxide radicals and the monohydrogen sulfate ion to various fragments of the carbohydrate chain. The latter preferentially ensued via 0,1A-, 0,2A-, 0,4A-, 0,2X-, 1,3X, or 2,4X-type ring scissions, using the MS nomenclature established for sugars (see labels in Figure 1).48 These fragmentation pathways are complementary to those documented for oligomeric saccharide subunits of GAGs in tandem MS48−50 and intact GAGs by source-induced fragmentation.35 Anion generation by pyrolysis DART is elucidated for heparin, DS, and OSCS in Figure 1. Heparin produced small sulfur-containing species (below m/z 100) in high abundance, and dermatan sulfate yielded GlcA, GalNAc, and related fragments. By contrast, the mass spectrum of OSCS was dominated by GalNac6S and associated breakdown products. Spectral differences are highlighted between DS and CSA in Figure S4. These results demonstrated that each GAG pyrolyzed in structurally distinct ways. Multivariate data analysis was developed to evaluate pyrolysis data and rapidly differentiate CSA, DS, heparin, and OSCS. A survey of the mass spectral features revealed that these GAGs
■
EXPERIMENTAL SECTION Chemicals. Sodium salts of CSA (≥∼70% pure with impurities of ∼10% DS and ∼20% chondroitin sulfate C; lot #060M1695 V) from bovine trachea, DS (≥90% pure; lot #080M1668 V), and heparin (grade I-A; lots #SLBB6758 V or #011M2040 V) from porcine intestinal mucosa, ammonium hydroxide (5 N), and LC MS-grade methanol and water were obtained from Sigma-Aldrich. Reference standards for heparin (lot #G0I116) and contaminated heparin (∼10% OSCS, lot #F0H211) were from the US Pharmacopeia. Standard OSCS was synthesized by sulfation of chondroitin A and characterized using SAX-LC, capillary electrophoresis, and NMR as detailed elsewhere.16 DART MS. The commercial electrospray source of an ion trap mass spectrometer (LCQ Fleet, ThermoScientific, Inc., West Palm Beach, FL) was replaced by a DART ion source (version 1.0, IonSense, Inc.). DART operating conditions were optimized as discussed in the Supporting Information (SI); the desorption gas was helium, and the temperature of analysis was 400 °C. Mass spectra were recorded between m/z 50−300 unless otherwise noted. Automatic gain control (AGC) was set at a threshold of 2,000. For tandem MS experiments, mass-selected ions were trapped and fragmented using helium collision gas and wide-band activation with 20−25 eV energies. Sample Measurement. Stock solutions of GAGs were prepared with water at 50 μg/μL concentration and diluted to 10−20 μg/μL in water for positive ion-mode or 1% ammonium hydroxide for negative ion-mode measurements. Solutions were kept in ice−water mix (∼4 °C) until measurement the same day. Glass capillaries (DipIt, IonSense) were cleaned by sonication in 50% methanol and air-dried and then utilized to transfer ∼150 nL of the cooled sample solution into the DART source (see the SI). Each sample was measured in 5−10 technical replicates with the same capillary, amounting to
