Pharmaceutical Applications Notebook

Anions • Cations • Organic Acids • Carbohydrates Table of Contents

Index of Analytes...... 3 Introduction to Pharmaceuticals...... 4 UltiMate 3000 UHPLC+ Systems...... 5 IC and RFIC Systems...... 6 MS Instruments...... 7 Chromeleon 7 Chromatography Data System Software...... 8 Process Analytical Systems and Software...... 9 Automated Sample Preparation...... 10

Analysis of Anions, Cations, Organic Acids, and Carbohydrates...... 11 Determination of Transition Metals in Serum and Whole Blood by Ion Chromatography ...... 12 Determination of Trifluoroacetic Acid (TFA) in Peptides...... 17 Quantification of Anions in Pharmaceuticals...... 23 Determination of Inorganic Anions and Organic Acids in Fermentation Broths...... 31 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography...... 42 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography...... 49 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer...... 56 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection...... 64 Quantification of Carbohydrates and Glycols in Pharmaceuticals...... 71 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths...... 79 Determination of Inorganic Counterions in Pharmaceutical Drugs Using Capillary IC...... 89 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column...... 91 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection...... 99 A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium...... 107

Column Selection Guide...... 113 Thermo Scientific Acclaim Column Selection Guide...... 114

Transferring HPLC Methods to UHPLC...... 115 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator...... 116 Index of Analytes

Anions...... 17, 23, 31, 42, 49, 64 Magnesium...... 89 Aromatic acids...... 17 Maltitol...... 71 Calcium...... 89 Mannitol...... 71, 79 Carbohydrates...... 71, 79 Organic acids...... 31 Chloride...... 89 Oversulfated chondroitin sulfate...... 99, 107 Cyanate...... 56 Oxalate...... 31 Dermatan sulfate...... 99, 107 Propylene glycol...... 71 Galactosamine...... 91 Sodium...... 89 Glucosamine...... 91 Sorbitol...... 71, 79 Glucose...... 71, 79 Sucrose...... 71, 79 Glycerol...... 71, 79 Transition metals...... 12 Glycols...... 71, 79 Trifluoroacetic acid...... 17 Heparin...... 91, 99, 107 Trifluoroacetate...... 42

3 Index of Analytes Introduction to Pharmaceuticals

The pharmaceutical industry is the largest consumer Although, some of the applications published in of high-performance liquid chromatography (HPLC) this notebook were created some time ago, they are instrumentation. In drug discovery, HPLC and ion still relevant today. In the event that specific models chromatography (IC) systems are used both as stand- of systems or modules used in these applications are alone tools and as front ends for mass spectrometers to no longer available, their methods may still be used on screen drug candidates. In pre-clinical development, they current instrumentation with similar performance. are used for analyzing in-vitro and in-vivo samples. In Thermo Scientific and Dionex Integrated Systems clinical trials, they are used to gather data on a potential drug’s safety and efficacy. They are used in manufacturing Dionex Products are now a part of the Thermo for many tasks including quality assurance/quality control Scientific brand, creating exciting new possibilities for (QA/QC), and the validation of cleaning procedures. scientific analysis. Now, leading capabilities in liquid This applications notebook has been compiled to chromatography (LC), IC, and sample preparation are help the pharmaceutical scientist by providing a together in one portfolio with those in mass spectrometry wide range of application examples relevant to the (MS). Combining Dionex’s leadership in chromatography pharmaceutical market. with Thermo Scientific’s leadership position in mass Thermo Fisher Scientific understands the demands spec, a new range of powerful and simplified workflow of chemical analysis in the pharmaceutical industry. solutions now becomes possible. Our separation and detection technologies, combined For more information on how the new line-up of with experience and applications competence, provide Thermo Scientific products can expand your capabilities solutions for the analysis of inorganic ions, small drug and provide the tools for new possibilities, choose one of molecules, and large components such as biologics and our integrated solutions: polysaccharides. Your laboratory now has a partner who • Ion Chromatography and Mass Spectrometry can help you conduct reliable, accurate, and fast analyses. • Liquid Chromatography and Mass Spectrometry This notebook contains a wide range of pharmaceutical- • Sample Preparation and Mass Spectrometry related application notes and relevant information that will help address your challenges in drug discovery, development, and manufacturing.

4 Introduction UltiMate 3000 UHPLC+ Systems

Best-in-class HPLC systems for all your chromatography needs

Thermo Scientific Dionex UltiMate 3000 UHPLC+ • Thermo Scientific DionexViper and nanoViper–the Systems provide excellent chromatographic performance first truly universal, fingertight fitting system even at while maintaining easy, reliable operation. The basic UHPLC pressures and standard analytical systems offer ultra HPLC Thermo Fisher Scientific is the only HPLC company (UHPLC) compatibility across all modules, ensuring uniquely focused on making UHPLC technology maximum performance for all users and all laboratories. available to all users, all laboratories, and for all analytes. Covering flow rates from 20 nL/min to 10 mL/min with an industry-leading range of pumping, sampling, and Rapid Separation LC Systems: The extended flow- detection modules, UltiMate™ 3000 UHPLC+ Systems pressure footprint of the RSLC system provides the provide solutions from nano to semipreparative, from performance for ultrafast high-resolution and conventional conventional LC to UHPLC. LC applications. • Superior chromatographic performance RSLCnano Systems: The Rapid Separation nano LC System (RSLCnano) provides the power for high- • UHPLC design philosophy throughout nano, resolution and fast chromatography in nano, capillary, and standard analytical, and rapid separation liquid micro LC. chromotography (RSLC) Standard LC Systems: Choose from a wide variety • 620 bar (9,000 psi) and 100 Hz data rate set a new of standard LC systems for demanding LC applications benchmark for basic and standard analytical systems at nano, capillary, micro, analytical, and semipreparative • RSLC systems go up to 1000 bar and data rates up to flow rates. 200 Hz Basic LC Systems: UltiMate 3000 Basic LC Systems • ×2 Dual System for increased productivity solutions are UHPLC compatible and provide reliable, high- in routine analysis performance solutions to fit your bench space and • Fully UHPLC compatible advanced chromatographic your budget. techniques

Liquid Chromatography Systems 5 IC and RFIC Systems

A complete range of ion chromatography solutions for all performance and price requirements

For ion analysis, nothing compares to a Thermo Dionex ICS-5000: Developed with flexibility, Fisher Scientific ion chromatography system. Whether modularity, and ease-of-use in mind, the Dionex you have just a few samples or a heavy workload, ICS-5000 combines the highest sensitivity whether your analytical task is simple or challenging, we with convenience have a solution to match your needs and budget. And with Dionex ICS-2100: An integrated Reagent-Free IC your IC purchase, you get more than just an instrument— (RFIC™) system for electrolytically generated isocratic you get a complete solution based on modern technology and gradient separations with conductivity detection, now and world-class support. with electrolytic sample preparation. Dionex ICS-1600: The Dionex ICS-1600 combines • Thermo Scientific Dionex ICS-5000:The world’s high sensitivity with convenience. Now ready for eluent first capillary IC system regeneration, with available dual-valve configuration for • Dionex ICS-2100: Award-winning integrated automated sample preparation. Reagent-Free™ IC system Dionex ICS-1100: With dual-piston pumping • Dionex ICS-1600: Standard integrated IC system and electrolytic suppression. Now ready for eluent • Dionex ICS-1100: Basic integrated IC system regeneration, with available dual-valve configuration for • Dionex ICS-900: Starter line IC system automated sample preparation. Ranging from the Dionex ICS-900 to the ICS-5000, Dionex ICS-900: Can routinely analyze multiple these IC systems cover the entire range of IC needs anions and cations in 10–15 min—fully automated with and budgets and come with superior support and Displacement Chemical Regeneration (DCR). service worldwide.

6 Ion Chromatography Systems MS Instruments

Single-point control and automation for improved ease-of-use in LC/MS and IC/MS

Thermo Fisher Scientific provides advanced • Thermo Scientific Dionex Chromeleon software for integrated IC/MS and LC/MS solutions with superior single-point method setup, instrument control, and ease-of-use and modest price and space requirements. data management UltiMate 3000 System Wellness technology and • Compatible with existing IC and LC methods automatic MS calibration allow continuous operation • The complete system includes the MSQ Plus™ with minimal maintenance. The Dionex ICS-5000 mass spectrometer, PC datasystem, electrospray instrument and the family of RFIC systems automatically ionization (ESI) and atmospheric pressure chemical remove mobile phase ions for effort-free transition to ionization (APCI) probe inlets, and vaccum system MS detection. You no longer need two software packages to operate • Thermo Scientific MSQ Plus mass spectrometer, the your LC/MS system. Chromeleon™ LC/MS software smallest and most sensitive single quadrupole on the provides single-software method setup and instrument market for LC and IC control; powerful UV, conductivity, and MS data analysis; • Self-cleaning ion source for low-maintenance and fully integrated reporting. operation

7 MS Instruments Chromeleon 7 Chromatography Data System Software

The fastest way to get from samples to results.

Discover Chromeleon software version 7, the Chromeleon software version 7 is a forward-looking chromatography software that streamlines your path solution to your long-term chromatography data needs. from samples to results. Get rich, intelligent functionality It is developed using the most modern software tools and and outstanding usability at the same time with technologies, and innovative features will continue to be Chromeleon software version 7—the Simply Intelligent™ added for many years to come. chromatography software. The Cobra™ integration wizard uses an advanced mathematical algorithm to define peaks. This ensures that • Enjoy a modern, intuitive user interface designed noise and shifting baselines are no longer a challenge around the principle of operational simplicity in difficult chromatograms. When peaks are not fully • Streamline laboratory processes and eliminate errors resolved, the SmartPeaks™ integration assistant visually with eWorkflows, which enable anyone to perform a displays integration options. Once a treatment is selected, complete analysis perfectly with just a few clicks the appropriate parameters are automatically included in • Access your instruments, data, and eWorkflows the processing method. instantly in the Chromeleon Console Chromeleon software version 7 ensures data • Locate and collate results quickly and easily using integrity and reliability with a suite of compliance tools. powerful built-in database query features Compliance tools provide sophisticated user management, • Interpret multiple chromatograms at a glance using protected database structures, and a detailed interactive MiniPlots audit trail and versioning system. • Find everything you need to view, analyze, and report data in the Chromatography Studio • Accelerate analyses and learn more from your data through dynamic, interactive displays • Deliver customized reports using the built-in Excel- compatible speadsheet

8 Chromatography Data Systems Process Analytical Systems and Software

Improve your process by improving your process monitoring with a Thermo Scientific Dionex on-line IC or HPLC system

Our process analytical systems provide timely results • The Thermo Scientific Integral Migration Path by moving liquid chromatography-based measurements approach lets you choose the systems that best meets on-line. Information from the Thermo Scientific Dionex your needs Integral process analyzer can help reduce process ™ variability, improve efficiency, and reduce downtime. The Integral Migration Path approach enables These systems provide comprehensive, precise, accurate on-line IC/HPLC to generate timely, high-resolution information faster than is possible with laboratory-based information when monitoring a small-scale reactor results. From the lab to the factory floor, your plant’s in a process R&D lab, in a pilot plant, or improving performance will benefit from the information provided current manufacturing plant processes. No matter what ™ by on-line LC. the application, the Integral process analyzer has the versatility to place a solution using on-line IC/HPLC, • Characterize your samples completely with whenever and wherever it is needed. multicomponent analysis • Reduce sample collection time and resources with Integral: The Integral Migration Path approach: automated multipoint sampling System solutions wherever you need them: lab, pilot plant, or manufacturing • Improve your process control with more Chromeleon Process Analytical (PA) Software: timely results Chromeleon PA software provides unique capabilities to • See more analytes with unique detection capabilities support on-line IC or HPLC analysis • 25 years of experience providing on-line IC and HPLC capabilities to a wide range of industries

9 Process Analytical Systems and Software Automated Sample Preparation

Better extractions in less time with less solvent

Solvent extractions that normally require labor- Dionex ASE systems are also used by intensive steps are automated or performed in minutes, government agencies: with reduced solvent consumption and reduced sample • US EPA Method 3545A handling using the Thermo Scientific Dionex Accelerated Solvent Extractor (ASE) system or • CLP SOW OLM 0.42 AutoTrace 280 Solid-Phase Extraction (SPE) instrument. • ASTM Standard Practice D7210 The Dionex ASE™ system is dramatically faster • Chinese Method GB/T 19649-2005 than Soxhlet, sonication, and other extraction methods, • German Method L00.00-34 and uses significantly less solvent and labor. Accelerated The Dionex AutoTrace™ system is an automated solvent extraction methods are accepted and established in SPE instrument for extractions of large volume liquid the environmental, pharmaceutical, foods, polymers, and sample matrixes. Dioenx AutoTrace systems automate the consumer product industries. standard SPE steps of condition, load, rinse and elute to reduce sample handling and improve productivity. Dionex AutoTrace systems are available in cartridge or disk formats.

10 Automated Sample Preparation Analysis of Anions, Cations, Organic Acids, and Carbohydrates Pharmaceutical Applications Notebook Application Note 108

Determination of Transition Metals in Serum and Whole Blood by Ion Chromatography

INTRODUCTION This application note describes an attractive alterna­ The determination of transition metals in physiological tive to traditional spectroscopic methods by using the fluids is of considerable interest in clinical chemistry. In principles of ion exchange. As a sample moves through recent years several studies have linked the concentra­ the ion exchange column, bands of transition metals tions of specific transition metals to various diseases. Low migrate through at differential rates determined largely serum copper level is used as a marker for Wilson’s dis­ by the relative affinities of the different metal-ligand ease. Serum copper levels are elevated in a large number complexes for the stationary ion exchange sites. A strong of chronic and acute illnesses such as Hodgkin’s disease, metal complex-ing colorimetric reagent is supplied pneu­ leukemia, and many other malignancies.1 Zinc is an im­ matically and mixed with the column effluent. The bands portant nutritive factor as well as a cofactor for many of transition metals are then determined at a visible wave­ metalloenzymes. Zinc is necessary for the growth and length using an absorbance detector.7 Separation between division of cells, especially during the stages of life when individual metals can be enhanced or altered simply by growth rates are high. Zinc deficiency is associated with changing eluents. Figure 1 illustrates the selectivity dif­ syndromes that cause short stature and dwarfism.2 There is ferences observed on an IonPac® CS5A column when us­ also interest in the biochemical relationship of copper and ing A) a pyridine-2,6-dicarboxylic acid eluent, and B) an zinc.3 Studies have linked an increase in plasma copper oxalic acid eluent. This method is precise, sensitive, and level with decreasing plasma zinc concentration in child­ requires mini­mum sample preparation. hood lymphatic leukemia. Determination of iron in whole blood is used to monitor anemia. REcommended eQUIPMENT Traditionally, atomic absorption spectrophoto­ Dionex DX-500 system consisting of: metric (AAS) techniques have been used by most clinical GP40 Gradient Pump chemistry laboratories to determine transition metals in AD20 Absorbance Detector physio-logical fluids. These techniques have their limita­ LC20 Chromatography Module tions. Flame AAS has limited sensitivity for copper, and PC10 Postcolumn Pneumatic Controller graphite atomic absorption spectrophotometry is susceptible PC10 Automation Kit (optional) to solute vaporization interferences such as depression of PeakNet® Chromatography Workstation element signal, especially in physiological samples.4–6 The protein content of the physiological fluid samples can cause absorption abnormalities. High sodium chloride content can hamper sensitivity, linearity, and cause burner clogging.

12 Determination of Transition Metals in Serum and Whole Blood by Ion Chromatography REAGENTS AND STANDARDS CONDITIONS Reagents Either of two analytical systems may be used for Deionized water, 17.8 MΩ-cm resistance or higher the chromatographic determination of transition metals. MetPac™ PDCA Eluent Concentrate and/or MetPac Ox­ The most current method employs the IonPac CS5A, a alic Acid Eluent Concentrate highly efficient, solvent-compatible, mixed anion/cation MetPac Postcolumn Diluent exchange column. An older column, the IonPac CS5, may be used for these analyses, but efficiencies are superior 4-(2-Pyridylazo)resorcinol, monosodium, monohydrate and cadmium is better resolved on the CS5A. Refer to (P/N 039672) Table 1 for a complete listing of experimental conditions Nitric acid, trace-metal grade (Fisher Scientific) for both columns. Sulfuric acid (Fisher Scientific) Hydrogen peroxide, 30% (Fisher Scientific) PREPARATION OF SOLUTIONS AND REAGENTS Trichloroacetic acid (Fluka Chemika-BioChemika) Two eluent systems can be used for transition metal separations with the IonPac CS5A or CS5 column. The Standards PDCA eluent is used for iron, copper, nickel, zinc, cobalt, Transition metal standards of 1000 mg/L are avail­ cadmium, and manganese. The oxalic acid eluent is used able from chemical supply companies (e.g. Aldrich, Sig­ for lead, copper, cobalt, zinc, and nickel. Cadmium and ma) for use with atomic absorption spectrometry. These manganese coelute using the oxalic acid eluent. are always dissolved in dilute acid solutions and can be used as IC standards. MetPac PDCA Eluent Dilute 200 mL of the MetPac PDCA Eluent Concen- trate to 1.0 L with deionized water.

Table 1. Conditions for Two Analytical Systems Used for the Chromatographic Determination of Transition Metals

Columns IonPac CS5A analytical and IonPac CG5A guard IonPac CS5 analytical and IonPac CG5 guard

Eluents A) MetPac PDCA Eluent (Alternatively, 8.0 mM PDCA, A) 6.0 mM PDCA 66 mM potassium hydroxide, 74 mM formic acid, and 40 mM Sodium hydroxide 5.6 mM potassium sulfate may be used.) 90 mM Acetic acid

B) MetPac Oxalic Acid Eluent (Alternatively, 8 mM oxalic B) 50 mM Oxalic acid acid, 50 mM potassium hydroxide, and 100 mM 95 mM Lithium hydroxide tetramethylammonium hydroxide may be used.)

Flow Rate 1.2 mL/min 1.0 mL/min

Detection Absorbance, 530 nm Absorbance, 520 nm

Postcolumn Reagent 0.5 mM PAR, dissolved in MetPac Postcolumn 0.4 mM PAR Diluent. (Alternatively, 1.0 M 2-dimethylaminoethanol, 1.0 M 2-dimethylaminoethanol 0.50 M ammonium hydroxide, and 0.30 M sodium 0.50 M Ammonium hydroxide bicarbonate may be used.) 0.30 M Sodium bicarbonate

Postcolumn Reagent Flow Rate 0.7 mL/min 0.5 mL/min

13 Determination of Transition Metals in Serum and Whole Blood by Ion Chromatography MetPac Oxalic Acid Eluent SAMPLE PREPARATION Dilute 100 mL of the MetPac Oxalic Acid Eluent Biological matrices contain higher concentrations of Concentrate to 1.0 L with deionized water. alkali and alkaline earth metals than transition metals. In such instances, a chelation concentration step can be used PAR [4-(2-Pyridylazo)resorcinol] Postcolumn Reagent where the alkali and alkaline earth metals are removed Prepare the postcolumn reagent directly in the from the matrix and the transition metals are selectively 1-L plastic reagent reservoir container. Add 0.15 g of concentrated. For the determination of trace metals in 4-(2-pyridy­lazo)resorcinol, monosodium, monohydrate, to physiological fluids or tissues, the sample must first be 1.0 L of the MetPac Postcolumn Diluent and ultrasonicate acid-digested to a single phase. In this application, the for five minutes. Add a stir bar and stir for several minutes serum and whole blood samples were digested using the to ensure that the PAR has completely dissolved. The col­ following procedure: to a 250-mL evaporation dish add or of the final solution should be yellow to yellow-orange. 10–100 mL of sample. Next, add 5 mL of concentrated Place the reagent container in the reagent reservoir. HNO3 and 2 mL of 30% H2O2. Evaporate on a hot plate at medium heat to a volume of 15 to 20 mL. Cover with a watch glass to avoid sample loss by spattering. Transfer A. PDCA Eluent Column: IonPac® CS5A, CG5A the concentrate and any precipitate to a 125 mL coni­ Eluent: MetPac® PDCA Eluent Flow Rate: 1.2 mL/min cal flask using 5 mL of concentrated HNO3. Add 10 mL Inj. Volume: 50 µL of concentrated H2SO4 and a few boiling chips or glass Detection: Absorbance, 530 nm with PAR in MetPac Postcolumn beads. Evaporate on a hot plate in a hood until dense Reagent Diluent white fumes of SO appear. If the solution does not clear, 0.2 Peaks: 1. Iron (III) 1.3 mg/L 3 2. Copper 1.3 add 10 mL of concentrated HNO3 and repeat evapora­ 3. Nickel 2.6 1 tion. Remove all HNO3 before continuing treatment. 7 4. Zinc 1.3 (All HNO is removed when the solution is clear and no 5 5. Cobalt 1.3 3 2 4 8 AU 6. Cadmium 6.0 brownish fumes are evident. ) Cool and dilute to about 3 6 7. Manganese 2.6 50 mL with eluent. 8. Iron (II) 1.3 8 Alternatively, samples can be digested using nitric acid. Add 40 g of concentrated nitric acid to approximate­ ly 75 g of sample. Add a 10-mL aliquot of the digested 0.0 0 2 4 6 8 10 12 14 11873 sample to 20 mL of 2 M ammonium acetate. The final pH Minutes of the sample should be 5.5. A trichloroacetic acid deproteinization procedure is B. Oxalate Eluent Column: IonPac CS5A, CG5A Eluent: MetPac Oxalic Acid Eluent sometimes used for serum and plasma samples. Add Flow Rate: 1.2 mL/min 0.2 mL of 50% trichloroacetic acid to 0.4 mL of serum or Inj. Volume: 50 µL Detection: Absorbance, 530 nm with PAR plasma sample. Centrifuge the mixture for 5 minutes at in MetPac Postcolumn 1500 x g. Inject an appropriate volume of the supernatant 0.2 3 Reagent Diluent (e.g. 25–50 L). Peaks: 1. Lead 5.0 mg/L µ 2. Copper 0.7 2 3. Cadmium 3.3 1 4. Cobalt 0.7 5 6 AU 5. Zinc 1.3 4 6. Nickel 2.0

0.0

024681012 Minutes 11871

Figure 1. Separation of transition metals using A) PDCA Eluent and B) Oxalate Eluent (IonPac CS5A).

14 Determination of Transition Metals in Serum and Whole Blood by Ion Chromatography DISCUSSION AND RESULTS The method outlined in this application note permits 0.12 1 Peaks: 1. Copper 0.87 mg/L 2. Zinc 0.63 rapid separation of various transition metals. The separa­ tions are based on one of two different eluent systems.

The first is a pyridine-2,6-dicarboxylic acid (PDCA) elu­ AU ent, which is a strong complexing agent that separates the metal ion complexes by anion exchange. PDCA is 2 best suited for iron(II) and iron(III), copper, nickel, zinc, cobalt, cadmium, and manganese (see Figure 2). This 0.0 method allows one to speciate the oxidation states of iron, 0 4 8 12 16 18 Fe(II) and Fe(III). However, since ferrous ion is easily Minutes 11731 oxidized to ferric, oxygen must be removed from the elu­ Figure 3. Determination of copper and zinc in serum ent by degassing. Oxygen should also be purged from the (IonPac CS5, Oxalic Acid Eluent). analytical column by pumping 0.1 M sodium sulfite

(12.6 g/L Na2SO3) through the column for 2 hours. The separated metals from the analytical column enter a postcolumn reaction system where they are deriva­ tized with 4-(2-pyridylazo)resorcinol and then detected at 0.1 1 Peaks: 1. Iron(III) 19.3 mg/L 2. Zinc 0.35 520–530 nm using a UV/visible absorbance detector. This method is ideal for complex matrices such as physi­ ological fluids. This method is highly sensitive, specific, AU and precise. Analyte recoveries for various physiological fluid matrices are listed in Tables 2 and 3.

2

0.0 Table 2. Precision and Recovery Data a 0 510 15 20 25 for Transition Metals in Whole Blood Minutes 11730

Figure 2. Determination of iron(III) and zinc in whole blood Analyte Amount Amount Mean RSD (IonPac CS5, PDCA eluent). Found Spiked Recovery (%) (mg/L) (mg/L) (%)

Iron(III)b 420 — — 2.7 An alternative eluent system uses an oxalic acid- Zinc 1.2 3.0 95 1.3 based eluent, which is a moderate strength complexing an = 7 replicates, 50 µL injected. agent that separates the metals by a mixed mode mecha­ bIron(III) was not spiked. nism. The oxalate eluent separates lead, copper, cobalt, zinc, and nickel (see Figure 3). Cadmium and manganese coelute with this eluent. Table 3. Precision and Accuracy Data for Transition Metals in Seruma

Analyte Amount Amount Mean RSD MDLb Found Spiked Recovery (%) µg/L (mg/L) (mg/L) (%)

Copper 0.57 1.0 98 3.8 45 Zinc 0.85 2.0 95 2.4 70

an = 7 replicates, 50 µL injected. b MDL = SD x (ts) 99%.

15 Determination of Transition Metals in Serum and Whole Blood by Ion Chromatography PRECAUTIONS SUPPLIERS The analytical flow path must have no metal com-po­ Aldrich Chemical Company nents. This includes tubing end fittings, stainless steel wash­ 1001 West Saint Paul Avenue ers, omni-fittings, etc., as well as columns and valves that P.O. Box 355 contain stainless steel. Use caution in preparing and transfer­ Milwaukee, WI 53233, USA ring reagents to minimize contamination. The prepared PAR Tel.: (800) 558-9160 is easily oxidized. If at anytime the PAR takes on a red color, Fisher Scientific, 711 Forbes Avenue it has been contaminated and should be discarded. Prepared Pittsburgh, Pennsylvania 15219-4785, USA reagents should be stored under an inert gas, such as nitrogen Tel.: (800) 766-7000 or helium, and used within two weeks of preparation. Be sure Fluka Chemika-BioChemika, Fluka Chemie AG that the eluent is being pumped through the columns when Industriestrasse 25, CH-9471 the postcolumn pneumatic controller is turned on. Buchs, Switzerland Failure to do so may cause the PAR reagent to back up Tel.: +81 755 25 11 through the analytical column. Sigma Chemical Company P.O. Box 14508 REFERENCES St. Louis, MO 63178, USA 1. Delves, F. E.; Alexander, F. W.; Lay, H. Brit. J. Hae- Tel.: (800) 325-3010 matol. 1985, 34, 101–107. 2. Wells, J. L.; James, D. K.; Luxton, R.; Pennock, C. A. Brit. Med. J. 1987, 294, 1054–1056. 3. Editorial, Nutr. Rev. 1984, 42, 184–186. 4. Nixon, D. E.; Moyer, T. P.; Johnson, P.; McCall, J.; Ness, A. B.; Fjerstad, W. H.; Wehde, M. B. Clin. Chem. 1986, 32, 1660–1665. 5. Delves, H. T. Prog. Anal. Atom. Spectrosc. 1981, 4, 1–48. 6. Passey, R. B.; Maluf, K. C.; Fuller, R. Anal. Biochem. 1985, 151, 462–465. 7. Rubin, R. B.; Heberling, S. S. Amer. Lab. 1987, 46–55. 8. Franson, M. H. (Ed.) “Standard Methods for the Examination of Water and Wastewater,” 15th Edition, Amer. Public Health Assoc., Washington, D. C., 1985.

16 Determination of Transition Metals in Serum and Whole Blood by Ion Chromatography Application Note 115

Determination of Trifluoroacetic Acid (TFA) in Peptides

Introduction cal column is passed through a suppressor that reduces the Trifluoroacetic acid (TFA) is commonly used in the total background conductance of the eluent and increases manufacturing process to release synthesized peptides the electrical conductance of the analyte ions. With sup­ from solid-phase resins. TFA or acetate is also used dur­ pressed conductivity, signal-to-noise ratios are improved ing reversed-phase HPLC purification of peptides. TFA approximately 50-fold compared to nonsuppressed is manufactured using acetate and fluoride as precursors, conductivity. and residual levels of these compounds may be present This application note describes the analysis of com­ whenever TFA is used. Residual TFA, fluoride, and, to mercially prepared, water-soluble peptides using ion a much lesser extent, acetate are toxic and undesirable in chromato-graphy. This method requires minimal sample peptides intended for preclinical and clinical studies. A preparation, and the analytes, fluoride, acetate, and method for the determination of TFA, acetate, and fluoride TFA, are easily separated without significant peptide must be suitable for peptide formulations and be capable interference. of verifying the removal of these anions during the pro­ duction process. EQUIPMENT TFA has been assayed by gas chromatography1–5, GC Dionex DX 500 system consisting of: mass spectroscopy6, reversed-phase HPLC7, isotacho- GP40 Gradient Pump phoresis8–10, infrared spectrometry11, titration12–13, spec­ CD20 Conductivity Detector or trophotometry14, and ion-exchange chromatography15–18. ED40 Electrochemical Detector Ion chromatography (IC) is advantageous because it is LC30 Oven or LC20 Chromatography Module sensitive, simple, and can be automated. AS3500 Autosampler The separation mechanism of IC is based on an PeakNet Chromatography Workstation anion-exchange displacement process occurring between the sample ions and eluent ions with the anion-exchange functional groups bonded to the stationary phase. A typi­ REAGENTS AND STANDARDS cal stationary phase consists of a grafted, solvent-com­ Reagents patible, alkyl-based ion-exchange resin. The separation of Sodium carbonate, 0.5 M (Dionex P/N 37162) TFA, acetate, and fluoride illustrated in this application Sodium bicarbonate, 0.5 M (Dionex P/N 37163) note uses a stationary phase functionalized with alkyl Deionized water, 18 ΜΩ-cm resistance or higher quaternary ammonium groups. Effluent from the analyti­

17 Determination of Trifluoroacetic Acid (TFA) in Peptides Standards SAMPLE PREPARATION Sodium fluoride, ACS grade (Fisher Scientific, Commercial Peptides Cat. No. S-299) Tyr-[Trp2]-MSH Release Inhibiting Factor;

Sodium acetate, trihydrate (Sigma Chemical Co., Tyr-Pro-Trp-Gly-NH2 Cat. No. S-8625) Trifluoroacetate salt, abbreviated here as MSH-RIF. Trifluoroacetic acid, anhydrous, protein sequencing grade [Sar1, Thr8]-Angiotensin II; (Sigma Chemical Co., Cat. No. T-1647) Sar-Arg-Val-Tyr-Ile-His-His-Pro-Thr Five Anion Standard (fluoride, chloride, nitrate, phos­ Acetate salt, abbreviated here as AT-II. phate, sulfate) (Dionex, P/N 37157) Ala-D-Isoglutaminyl-Lys-D-Ala-D-Lys Acetate salt, abbreviated here as IGA. CONDITIONS [Gln4]-Neurotensin; pGlu-Leu-Tyr-Gln-Asn- Columns: IonPac® AS14, Analytical (4 mm) Lys-Pro-Arg-Arg-Pro-Tyr-Ile-Leu IonPac AG14, Guard (4 mm) No counterion specified, abbreviated here as NT. Eluent: 3.5 mM Sodium carbonate/ FMRF Amide Related Peptide; 0.8 mM Sodium bicarbonate Asn-Arg-Asn-Phe-Leu-Arg-Phe amide Flow Rate: 1.2 mL/minute Trifluoroacetate salt, abbreviated here as FMRF. Inj. Volume: 10 µL Peptide Samples from In-Process Manufacturing Detection: Suppressed conductivity, ASRS™ Eleven Amino Acid Crude Peptide; Ac-D-Ala-Gly-Arg- AutoSuppression™ recycle and His-Tyr-Ala-Arg-Val-Ala-Leu-Arg-amide external water modes No purification, abbreviated as “Crude Peptide”. Expected Eleven Amino Acid >70% Pure Peptide; Ac-D-Ala-Gly- Background Arg-His-Tyr-Ala-Arg-Val-Ala-Leu-Arg-amide Conductivity: 15 µS Purified by gel permeation chromatography, Expected abbreviated as “GPC Pure Peptide”. System Operating Commercial peptide samples are dissolved in eluent Backpressure: 12.4 MPa (1800 psi) to peptide concentrations of 1 mg/mL. Peptide samples from in-process manufacturing are dissolved in eluent to PREPARATION OF SOLUTIONS AND REAGENTS dry weight concentrations of 1 mg/mL. Both the commer­ 3.5 mM Sodium carbonate / 0.8 mM Sodium bicarbonate cial and in-process peptide solutions are further diluted Combine 1980 mL of deionized water with 14.0 mL with eluent to concentrations of 40 and 400 µg/mL. of 0.5 M sodium carbonate and 3.2 mL of 0.5 M sodium Peptide solutions are also diluted with standard solutions bicarbonate. Degas for 20 minutes. Connect the eluent to evaluate the spike recovery by the method of standard reservoir to the instrument and pressurize with helium. addition. STOCK STANDARDS DISCUSSION AND RESULTS Prepare a 4.5-mg/mL stock analyte standard of fluo­ Figure 1 shows the separation of fluoride, acetate, ride by combining 10.0 mg of sodium fluoride with chloride, nitrate, phosphate, sulfate, and TFA using the 1.00 mL of water. Prepare a 7.2-mg/mL acetate standard 3.5 mM sodium carbonate/0.8 mM sodium bicarbonate by combining 10.0 mg of sodium acetate trihydrate with eluent. 1.00 mL of water. For a 9.9-mg/mL trifluoroacetate stock solution, mix 10.0 mg of trifluoroacetic acid with 1.00 mL of water. Combine and dilute standard solutions to de­ sired concentrations using the mobile phase eluent as the diluent. Standard solutions should be frozen until needed.

18 Determination of Trifluoroacetic Acid (TFA) in Peptides Table 1. Detection Limits 6 Peaks: 1. Fluoride Sample Mode Recycle Mode External Water 6 2. Acetate 1 3. Chloride (ng) (ng/mL) (ng) (ng/mL) 3 4. Nitrate Fluoride 0.3 30 0.1 10 4 5. Phosphate µS 6. Sulfate 7. Trifluoroacetate Acetate 2 200 1 100 5 7 Trifluoroacetate 6 600 3 300 2

2 Linearity 012 4 86 10 214 16 18 20

Minutes 13029 Fluoride standards of 0.023, 0.045, 0.23, 0.45, 2.3, 4.5, 23, 45, 230, 450 µg/mL were injected (n = 6 per con­ Figure 1. Seven common anions found in peptide samples. centration) for this study. The method was found to be lin­ ear for fluoride over the range tested (r2 = 1.000). Acetate Method Detection Limits and TFA were also linear (r2 = 0.999 in each case), using The method detection limits (MDL) for a 10-µL 0.36, 0.72, 3.6, 7.2, 36, 72, 360 µg/mL acetate standards injection of fluoride, acetate, and trifluoroacetate are vengi and 0.50, 1.0, 5.0, 100, 50, 100, 500 µg/mL trifluoroac­ in Table 1. The MDL is defined as the minimum concen­ etate standards (n = 6). For all three analytes, linearity was tration required to produce a signal-to-noise ratio of 3. demonstrated over at least three orders of magnitude. Two modes of suppression were compared for their effect on MDL. The recycle mode, which used the postdetector Stability eluent as feed through the regenerant chamber to achieve Standards of fluoride (4.5 µg/mL), acetate suppression, was compared to the external water mode, (7.2 µg/mL), and TFA (10 µg/mL) were injected over which used deionized water from a separate, pressurized 48 hours using an equilibrated system. Sample vials were bottle as regenerant feed. Suppression in the external at ambient temperature. Peak areas (Figure 2) and reten­ water mode compared to the recycle mode produces lower tion times (Figure 3) were reasonably stable throughout background noise, thus a lower MDL. The MDL can this period. be further decreased by increasing the injection volume above the 10 µL used in this Application Note.

250000 Fluoride

200000

150000 Area Units 100000 TFA

50000

Acetate

0 0510 15 20 25 30 35 40 45 50

Time (Hours) 13030

Figure 2. Peak area stability over 48 hours.

19 Determination of Trifluoroacetic Acid (TFA) in Peptides 14

TFA 12

10

8

6 Retention Time (Minutes) Retention Time

4 Acetate

2 Fluoride

0 0 510215 025 30 35 40 45 50

Time (Hours) 13031

Figure 3. Retention time stability over 48 hours.

Table 2. Recovery of Fluoride, Acetate, and Table 3. Effect of Peptide (FMRF) Concentration Trifluoroacetate from Peptides on Recovery of Fluoride, Acetate, and Trifluoroacetate

Percent Recovery (Mean) Percent Recovery (Mean)

Anion Spiked MSH-RIF AT-II IGA NT FMRF Peptide Inj. (ng) Fluoride Acetate TFA (µg/mL) 140 97.1 115 99.3 Fluoride 1.8 93.3 93.7 96.7 95.4 93.6 420 96.4 101 101 3.6 94.8 95.2 98.4 96.9 96.4 1010 97.7 102 120 5.4 96.4 97.4 99.4 98.0 99.9

Acetate 2.8 97.9 79.2 96.4 95.2 101 5.7 102 99.6 99.9 98.2 101 Recovery from Peptide Matrix 8.6 98.3 99.4 97.7 100 106 Table 2 shows the recovery of fluoride, acetate, and TFA 4.0 105 98.1 91.1 104 101 TFA from commercial peptides. Anions were spiked into 8.0 105 102 94.4 103 101 the peptide solutions by the method of standard addition. 12.0 98.3 100 99.7 106 103 Recovery for fluoride ranged from 93 to 100%, acetate from 79 to 106%, and TFA from 91 to 106%. Only fluo­ Precision ride showed a slightly higher recovery with increasing lev­ Precision is affected by concentration; RSD values els of standard. The amount of peptide (FMRF) injected increase as the concentrations approach the MDL. The had no effect on recovery (see Table 3). peak area RSD values for fluoride (45 ng/injection), acetate (72 ng/injection), and TFA (99 ng/injection) were 0.5, 2.3, and 3.8% respectively for 12 injections. Retention time precision (RSD) values were 0.4, 0.4, and 0.3% for fluoride, acetate, and TFA, respectively.

20 Determination of Trifluoroacetic Acid (TFA) in Peptides Table 4. Fluoride, Acetate, and Trifluoroacetate –1.5 in Commercial Peptides A Peaks: 1. Chloride 2. Sulfate mg anion/g peptide 3. TFA 3 Peptide Fluoride Acetate Trifluoroacetate µS MSH-RIF 0.26 0.90 202 AT-II 0.18 93 <0.3 2 1 IGA 24 80 <0.3 NT 0.02 0.13 184 –1.9 FMRF 0.02 0.15 193 018642 10 214161820 –1.5 13032 B Peaks: 1. Chloride Residual Anions in Commercial Peptides 2. Sulfate 3. TFA Commercially available peptides contained counteri­ 3 µS 1 ons that were quantifiable by ion chromatography. Table 4 2 lists the fluoride, acetate, and trifluoroacetate measured in the commercial peptides. All peptides labeled by the man­ ufacturer as trifluoroacetate salts (MSH-RIF and FMRF) contained TFA at approximately the same concentrations –1.9 (193–202 mg of TFA per gram of peptide). All peptides 018642 10 214161820 Minutes labeled as acetate salts (AT-II and IGA) contained acetate 13033 at measurable levels. The peptide NT was labeled as Figure 4. (A) Crude in-process peptide (10 µL of 40 µg/mL). containing salts, but did not indicate which salts. By this (B) GPC-purified in-process peptide (10 µL of 40 µg/mL). method, NT was determined to contain TFA at levels similar to those found in MSH-RIF and FMRF. Further­ more, IGA was labeled as containing acetate as the coun­ chloride and 2.6% sulfate). The GPC-purified peptide terion, but additionally contained fluoride. All the peptides sample also contained fluoride and nitrate at 0.026% and investigated in this Application Note also contained other 0.014% respectively. Knowing the amount and type of anions, such as chloride, sulfate, phosphate, and nitrate, these counterions in peptide preparations is important to which can be effectively resolved by this method. Other the ultimate safety and effectiveness of the product. Ion anions were detected but not identified. chromatography is therefore an effective in-process qual­ Residual Anions During In-Process Peptide Manufacturing ity control method. This ion chromatography method can This IC method can be used to evaluate the effec­ also assist with defining the mass balance of the peptide tiveness of purification during manufacturing. A crude preparations. synthetic peptide sample (prior to purification) was determined to contain 19.6% TFA by dry weight. Gel CONCLUSION permeation chromatography (GPC) was used as a pri­ This isocratic IC method using the IonPac AS14 mary purification step after synthesis, producing a peptide column with suppressed conductivity detection can be sample containing 16.7% TFA­­. Figures 4A and 4B show used to evaluate peptides for residual TFA, fluoride, and chromatograms of these peptide solutions. These results acetate. The method also resolves other common anions suggest that TFA was not effectively removed by GPC. such as chloride, sulfate, nitrate, and phosphate. TFA, Appreciable levels of chloride and sulfate were mea­ fluoride, and acetate can be detected at the mg/L level. sured in the crude peptide (0.02 and 1.0%, respectively). The recovery of fluoride, acetate, and TFA from peptide After GPC, much higher levels were observed (0.83% matrices is normally greater than 90%.

21 Determination of Trifluoroacetic Acid (TFA) in Peptides REFERENCES 11. Gossler, K.; Schaller, K. H.; Essing, H. G.; 1. Both, D. A.; Jemal, M. J. Chromatogr. 1992, 596, Fresenius, Z. Anal. Chem. 1976, 279, 112–113. 85–90. 12. Spirina, R. I.; Lyakhova, K. V. Zavod. Lab. 1989, 2. Mariorino, R. M.; Gandolfi,A. J.; Sipes, I. G. J. Anal. 55 (4), 6–7. Toxicol. 1980, 4, 250–254. 13. Fan, B.; Zhu, M.; Ma, Y. Huaxue Shijie. 1983, 24, 3. Dimitrieva, T. M.; Koemets, L. A. Prod. Khim. Prom- 11–13. sti. 1979, 2, 9–11. 14. Ilcheva, L.; Todorova, G. Acta Chim. Acad. Sci. 4. Witte, L.; Nau, H.; Fuhrhop, J. H. J. Chromatogr. Hung. 1979, 102, 113–120. 1977, 143, 329–334. 15. Simonzadeh, N. J. Chromatogr. 1993, 634, 125–128. 5. Karashima, D.; Shigematsu, A.; Furukawa, T.; Na­ 16. Kawaguchi, R.; Fujii, K.; Morio, M.; Yuge, O.; Hos­ gayoshi, T.; Matsumoto, I. J. Chromatogr. 1977, 130, sain, M. D. J. Med. Sci. 1989, 38, 27–34. 77–86. 17. Nakazawa, H.; Nagase, M.; Onuma, T. Bunseki Ka- 6. Gruenke, D. L.; Waskell, L. A. Biomed. Environ. gaku 1987, 36, 396–398. Mass Spectrom. 1988, 17, 471–475. 18. Fujii, K.; Morio, M.; Kikuchi, H.; Takiyama, R.; 7. Imbenotte, M.; Brice, A.; Erb, F.; Haguenoer, J. M. Katayama, T. Masui to Sosei 1984, 20 (Suppl.), 5–8. Talanta 1984, 31, 147–149. 8. Hirokawa, T.; Takemi, H.; Riso, Y. J.; Takiyama, R.; LIST OF SUPPLIERS Morio, M.; Fujii, K.; Kikuchi, H. J. Chromatogr. Fisher Scientific, 711 Forbes Ave., Pittsburgh, 1984, 305, 429–437. Pennsylvania, 15219-4785, U.S.A., 800-766-7000. 9. Janssen, P. S. L.; van Nispen, J. W. J. Chromatogr. Sigma Chemical Corporation, P.O. Box 14508, St. Louis, 1984, 287, 166–175. Missouri, 63178, U.S.A., 800-325-3010. 10. Morio, M.; Fujii, K.; Takiyama, R.; Chikasue, F.; Kikuchi, H.; Ribaric, L. Anesthesiology 1980, 52, 56–59.

22 Determination of Trifluoroacetic Acid (TFA) in Peptides Application Note 116

Quantification of Anions in Pharmaceuticals

INTRODUCTION In the methods outlined in this Note, the selec­ The United States Food and Drug Administration tivities of the IonPac AS14 and AS11 columns for (U.S. FDA)1–3 and regulatory agencies in other countries the analysis of anionic ingredients in pharmaceutical require that pharmaceutical products be tested for com­ formulations were compared. The AS14 packing has position to verify their identity, strength, quality, and pu­ a highly crosslinked core with an anion-exchange rity, with increased attention to inactive as well as active layer grafted to the surface. The anion-exchange ingredients. Analytical techniques that can adequately layer is functionalized with alkyl quaternary ammo­ test complex formulations composed of chromophoric nium functional groups and is grafted to crosslinked and nonchromophoric ingredients are desirable. Non- ethylvinylbenzene. This anion-exchange resin is se­ chromophoric ingredients, many of which are ionic, lective for the more hydrophobic anions. The AS11 cannot be detected by absorbance, but can be detected column packing has a pellicular structure consisting using suppressed conductivity. Suppressed conductiv­ of an alkyl quaternary ammonium latex bonded to ity is a powerful detection technique with a broad linear a crosslinked ethylvinylbenzene core. Another im­ range and very low detection limits. Suppression lowers portant difference between the two columns is that the background conductivity caused by the eluent and the AS14 column is designed for isocratic condi­ effectively increases the conductivity of the analyte.4–5 tions, while the AS11 can be used with hydroxide This Application Note describes the use of two gradients. Both columns are compatible with eluents anion-exchange columns with suppressed conductivity containing organic solvents, which can be used to detection to analyze common anions in pharmaceutical reduce undesirable secondary interactions some or­ formulations. Two oral, over-the-counter medications ganic anions have with stationary phases. Expected were selected as representative pharmaceutical products: detection limits, linearity, selectivity, accuracy, and a cough suppressant and a multisymptom cold/flu medi­ precision are reported for the AS11 column. cation. These formulations contain complex mixtures of ingredients that are commonly found in other medica­ EQUIPMENT tions, many of which are ionic and nonchromophoric. Dionex DX-500 system consisting of: Furthermore, these formulations also contain sugar GP40 Gradient Pump, with degas option alcohol, glycol, and carbohydrate ingredients that can CD20 Conductivity Detector or be analyzed using the IonPac® ICE-AS1, CarboPac™ ED40 Electrochemical Detector MA1, and CarboPac PA10 columns with electrochemical LC30 or LC25 Chromatography Ovens or 6 detection. LC20 Chromatography Module AS3500 Autosampler PeakNet Chromatography Workstation

23 Quantification ofAnions in Pharmaceuticals REAGENTS AND STANDARDS Eluent: Linear sodium hydroxide gradients: Reagents 0.5 mM Sodium hydroxide for 2.5 min, Sodium carbonate, 0.5 M (Dionex P/N 37162) then 0.5 to 5 mM sodium hydroxide Sodium bicarbonate, 0.5 M (Dionex P/N 37163) for 3.5 min, then 5 to 38 mM sodium hydroxide for 12 min, then 0.5 mM Sodium hydroxide, 50% (w/w) (Fisher Scientific, sodium hydroxide for 7 min. P/N SS254-500) Flow Rate: 2.0 mL/min Deionized water, 18 MΩ-cm resistance or higher (Prior to use, Dionex recommends testing the water for Injection Vol.: 10 µL trace anions using the intended ion chromatography Detection mode: Suppressed conductivity, ASRS, method.) AutoSuppression recycle mode Expected Standards Background Five Anion Standard (fluoride, chloride, nitrate, phos­ Conductivity: 0.5 µS phate, sulfate; Dionex P/N 37157) Expected System Sodium acetate, anhydrous (Fluka Biochemika, Operating P/N 71179) Backpressure: 12.4 MPa (1800 psi) Sodium bromide (Aldrich Chemical Co., P/N 31,050-6) Sodium chloride (Fisher Scientific, P/N S271-500) PREPARATION OF SOLUTIONS AND REAGENTS Sodium carbonate/bicarbonate eluent Sodium citrate, monohydrate (Fisher Scientific, 3.5 mM Sodium carbonate/0.8 mM Sodium P/N A104-500) bicarbonate Sodium benzoate (Sigma Chemical Co., P/N B-3375) Combine 1980 mL of deionized water with 14.0 mL Saccharin (Sodium salt; Sigma Chemical Co., of 0.5 M sodium carbonate and 3.2 mL of 0.5 M sodium P/N S-1002) bicarbonate. Degas for 20 minutes. Connect the eluent reservoir to the instrument and pressurize with helium. CONDITIONS System 1 Sodium hydroxide eluents Column: IonPac AS14 Analytical 5 mM Sodium hydroxide (P/N 46124), IonPac AG14 Guard Degas 2000 mL of deionized water for 20 min and (P/N 46134) combine with 520 µL of 50% (w/w) sodium hydroxide. Eluent: 3.5 mM Sodium carbonate 0.8 mM Sodium bicarbonate 100 mM Sodium hydroxide Degas 1990 mL of deionized water for 20 min and Flow Rate: 1.2 mL/min combine with 10.4 mL of 50% (w/w) sodium hydroxide. Injection Vol.: 10 µL Detection mode: Suppressed conductivity, ASRS™, STOCK STANDARDS AutoSuppression™ recycle mode Combine the Five Anion Standard (fluoride, chlo­ Expected ride, nitrate, phosphate, sulfate; Dionex P/N 37157) with Background acetate, bromide, citrate, benzoate, and saccharin stan­ Conductivity: 15 µS dards and purified water to yield stock concentrations of: Expected System Operating Fluoride 1.6 mg/L Phosphate 12 mg/L Backpressure: 12.4 MPa (1800 psi) Acetate 9.8 mg/L Chloride 2.4 mg/L Nitrate 8.0 mg/L Citrate 10 mg/L System 2 Bromide 9.8 mg/L Benzoate 10 mg/L Column: IonPac AS11 Analytical Sulfate 12.0 mg/L Saccharin 100 mg/L (P/N 44076), IonPac AG11 Guard (P/N 44078), ATC-1 Anion Trap Column (P/N 37151) 24 Quantification ofAnions in Pharmaceuticals Dilute the stock solution with water to the desired Table 2. Multisymptom Cold/Flu Medication concentrations. Pseudoephedrine Hydrochloride Active For determinations of linear range, combine 10-g/L Acetaminophen Active solutions of chloride, bromide, benzoate, citrate, and Dextromethorphan Hydrobromide Active saccharin to make a 1.0-g/L solution of standard mix. Dilute with water to concentrations of 0.8, 0.6, 0.4, 0.2, Citric Acid Inactive 0.1, 0.08, 0.06, 0.04, 0.02, 0.01, 0.005, and 0.001 g/L. FD&C Yellow #6 Inactive Dilute saccharin and citrate separately to evaluate these Flavor Inactive analytes without interference using concentrations of Glycerin (glycerol) Inactive 1.0, 0.8, 0.6, 0.4, 0.2, and 0.1 g/L. Polyethylene Glycol Inactive Propylene Glycol Inactive SAMPLE PREPARATION Purified Water Inactive Dilute viscous products with water on a weight per Saccharin Sodium Inactive weight basis. A 10-fold (w/w) dilution was obtained by Sodium Citrate Inactive combining 1 gram of medication with 9 grams of water. Sucrose Inactive The multisymptom cold/flu medication was further di­ luted to a 100-fold (w/w) final concentration. Determine the densities of the products by measuring the weights Any purified water used for dilutions should be of known volumes. Calculate the final concentrations of tested for trace anions by ion chromatography prior to the ingredients based on the densities of these medica­ use. Sample containers should be tested for residual tions. The ingredients of each medication are presented anions prior to use by adding pure water, shaking or in Tables 1 and 2. The ingredients noted in bold-face vortexing, and then testing the liquid. Plastic vials are type can be analyzed by anion-exchange columns with usually lower in residual anions than glass. Prerinsing suppressed conductivity; other ingredients listed can be the vials with purified water can reduce artifacts. analyzed using CarboPac columns with electrochemical detection.6 DISCUSSION AND RESULTS Selectivity Figure 1 shows the separation of fluoride, acetate, Table 1. Cough Suppressant chloride, nitrate, phosphate, and sulfate standards using Dextromethorphan Hydrobromide Active a 3.5 mM sodium carbonate/0.8 mM sodium bicarbon­ Citric Acid Inactive ate eluent with the IonPac AS14 column. The isocratic FD&C Red 40 Inactive conditions eliminate any need to reequilibrate the Flavors Inactive column, thereby decreasing run times and increasing Glycerin (glycerol) Inactive throughput. These anions can be analyzed within Propylene Glycol Inactive 14 minutes. Citrate, benzoate, and saccharin are not eluted by this method. Saccharin Sodium Inactive Sodium Benzoate Inactive Sorbitol Inactive Water Inactive

25 Quantification ofAnions in Pharmaceuticals Saccharin retention times significantly shorten Column: IonPac AG14, AS14 Peaks: 1. Fluoride 4 mg/L Eluent: 3.5 mM Sodium carbonate/ 2. Acetate 15 above 1000 ng per injection, causing coelution with cit­ 0.8 mM Sodium bicarbonate 3. Chloride 6 Flow Rate: 1.2 mL/min 4. Nitrite 10 rate. Saccharin retention depends on secondary interac­ Inj. Volume: 10 µL 5. Bromide 15 tions with the column. At high concentrations, saccharin Detection: Suppressed conductivity, ASRS, 6. Nitrate 20 AutoSuppression recycle mode 7. Phosphate 30 4 exceeds the capacity of the column and may elute earlier 8. Sulfate 30 1 8 than at lower concentrations. When both saccharin and 6 3 citrate coexist in a formulation, they both should be ad-

µS 5 justed to below 100 mg/L for a 10-µL injection. Saccha­ 7 4 rin peaks tail, but the addition of organic solvents in the 2 eluent (e.g., 10% acetonitrile) reduces tailing. Addition 1 of acetonitrile alters the column selectivity, decreases peak area response, and increases background noise. 02468101214 Minutes 12743 Although not presented here, nearly a dozen over- the-counter medications have been analyzed using Figure 1. Rapid separation of inorganic anions. the IonPac AS14 and AS11 columns with suppressed conductivity. These medications include both solid and liquid formulations such as nasal and oral decongestants, With the IonPac AS11 column, inorganic anions, ac­ astringents, antacids, enemas, sleep aids, analgesics, etate, citrate, benzoate, and saccharin are eluted using a cleaning and disinfecting solutions, antihistamines, and sodium hydroxide gradient (see Figure 2). Although the allergy syrups. The known anionic ingredients in each gradient results in slightly longer run times, both organic formulation were separated from each other without any and inorganic anions are effectively separated. Shorter apparent interference. Trace levels of unlabeled ingredi­ run times are possible for either method by adjusting ents were also detected as minor peaks and, when identi­ the eluent strength, but some resolution may be lost for fied, measured at low concentrations. fluoride and acetate. Method Detection Limits The method detection limits (MDL) for a 10-µL in­ Column: IonPac AG11, AS11 Peaks: 1. Fluoride 2 mg/L jection of common pharmaceutical anions with the AS11 Eluent: 0.5 mM Sodium hydroxide, hold 2. Acetate 10 for 2.5 min; 0.5 mM to 5 mM Sodium 3. Chloride 2 column are described in Table 3. The MDL is defined as hydroxide in 3.5 min; 5 mM to 38 mM 4. Bromide 10 the minimum concentration required to produce a signal- Sodium hydroxide in 12 min 5. Nitrate 8 Flow Rate: 2 mL/min 6. Benzoate 10 to-noise ratio of 3. The MDL can be further decreased Inj. Volume: 10 µL 7. Sulfate 12 Detection: Suppressed conductivity, ASRS, 8. Phosphate 12 by increasing the injection volume above the 10 µL used AutoSuppression recycle mode 9. Citrate 10 in this Application Note or by using external water mode 8 7 10. Saccharin 100 suppression.7

µS 45 8 10 12 3 9 6 Table 3 . Estimated Lower Detection Limits

0 System 2 (IonPac AS11 column) 02468101214161820 ng µg/L Minutes 12744 Fluoride 0.3 30 Figure 2. Gradient separation of both inorganic and organic Acetate 2 200 anions. Chloride 0.5 50 Bromide 4 400 Nitrate 3 300 Sulfate 1 100 Phosphate 4 400 Citrate 4 400 Benzoate 7 700 Saccharin 20 2000

26 Quantification ofAnions in Pharmaceuticals Linearity (10 ng to 10,000 ng), the correlation coefficient for Chloride, bromide, benzoate, citrate, and saccharin benzoate was 0.995. Although linearity is good for standards ranging from 1 to 1000 mg/L (10 ng to citrate and saccharin when run independently, the two 10,000 ng) were injected (in triplicate) on the AS11 peaks coelute at concentrations above 100 mg/L column (data not shown). The peak area response was (1000 ng). Therefore, linearity cannot be evaluated when found to be linear for chloride, bromide, citrate, and sac­ both compounds are present at this concentration. For charin over this range (r2 ≥ 0.999). Benzoate was linear all analytes, linearity was demonstrated over at least two over the range of 1 to 200 mg/L (10 ng to 2,000 ng per orders of magnitude. injection; r2 = 0.999). For the range of 1 to 1000 mg/L Representative calibration curves for both the AS14 and the AS11 columns are presented in Figures 3 and 4.

12 Fluoride r2 = 0.9996

2 10 10-µL Injection Chloride r = 0.9988 Sulfate r2 = 0.9998

2 ) 8 Nitrate r = 0.9991 5 Bromide r2 = 0.9996 6 Phosphate r2 = 0.9999

Area Units (x10 Acetate r2 = 0.9976 4

2

0 050 100 150 200 250 ng Injected 12573

Figure 3. Method linearity for the IonPac AS14 column (isocratic method).

8

7 Fluoride r2 = 0.9999 10-µL Injected Chloride r2 = 0.9999 6 Sulfate r2 = 0.9998

2

) 5 Nitrate r = 0.9997 5 Bromide r2 = 0.9998 4 Acetate r2 = 0.9991

2 Area Units (x10 3 Phosphate r = 0.9999 Citrate r2 = 0.9998 2 Benzoate r2 = 0.9999 Saccharin r2 = 0.9990 1

0 0204060 80 100 120 12754 ng Injected

Figure 4. Method linearity for the IonPac AS11 column (gradient method).

27 Quantification ofAnions in Pharmaceuticals Table 4. Peak Area and Retention Time Precision* Table 5. Peak Area and Precision of Pharmaceutical Products* RSD (%) Area Retention Time RSD (%) Measured Cough Cold/Flu Chloride 0.1 0.2 Analyte Conc. (mg/L) Suppressant Medication Bromide 0.1 0.1 Benzoate 0.2 0.1 Bromide 30 0.5 – Citrate 0.3 0.4 Bromide 1 – 1.0 Saccharin 0.4 0.3 Chloride 3 – 0.3 * 100-mg/L standards, 10 µL per injection (n = 15). * 10-fold (w/w) dilutions of cough suppressant, 100-fold (w/w) dilutions of cold/flu medication, 10-µL injection (n = 10).

Precision The peak area and retention time RSDs using the Table 6. Recovery of Active Ingredient AS11 column are presented in Table 4. Area precision Counterions in Medications is affected by concentration; RSD values increase as the concentrations approach the MDL. Dextromethorphan Pseudoephedrine Replicate, 10-µL injections (n = 10) of a 10-fold hydrobromide hydrochloride (w/w) dilution of cough suppressant yielded a bromide Cough suppressant 98.9% N/A peak area RSD of 0.5% when measured at 30 µg/mL. Multisymptom cold/flu 97.3% 109% A 100-fold (w/w) dilution of the multisymptom cold/ formulation flu medication resulted in a bromide RSD value of 1.0% when measured at 1 µg/mL for 10-µL injections; active ingredient in a pharmaceutical can be a valu­ chloride was 0.3% at 3 µg/mL for 10-µL injections. The able troubleshooting tool. For example, during stability retention time RSD values ranged from 0.1% to 0.2% studies, samples can either change in concentration or for these anions. These sample RSDs are summarized in components can decompose. Analysis of both the or­ Table 5 and are consistent with those found for standards ganic component and the inorganic counterion can serve analyzed at equivalent concentrations. to differentiate between these symptoms. The dosage of inactive ingredients is not specified Recovery from Sample Matrix on the label of most medications. Therefore, it was not Figures 5 and 6 show the separation of bromide possible to evaluate the accuracy of the formulation for from other ingredients in cough suppressant using the benzoate, citrate, and saccharin. However, these ingredi­ AS14 and AS11 columns, respectively. Bromide is the ents were quantified (see Figures 6 and 8). counterion in the active ingredient dextromethorphan hydrobromide. Figures 7 and 8 show the separation of both bromide and chloride from other ingredients in Other Anions in Pharmaceutical Products a multisymptom cold/flu formulation. Chloride is the Besides the labeled content of the pharmaceutical counterion in pseudoephedrine hydrochloride. These products, other anionic ingredients may also be present. anions are unique to these active ingredients in both These may arise from trace levels of ingredients found in medications; therefore, their quantification is an orthogo­ the raw materials used in manufacturing. Expanding the nal method for the determinations of dextromethorphan baseline of the above chromatograms (Figures 6B and and pseudoephedrine in these formulations. 8B) reveals the presence of minor peaks. Some can be The measured levels of these anions using the AS11 identified based on retention time, while others cannot. column are compared to the labeled values of these Sulfate (2 µg/mL) and phosphate (1 µg/mL) were active ingredients in each formulation and expressed as measured in the cough suppressant. Five other minor a percent recovery (see Table 6). In all cases, the levels peaks were observed but were not identified. Sulfate of bromide and chloride were within 10% of the label (13 µg/mL) was measured in the multisymptom cold/flu value. This orthogonal method for quantification of an formulation, while three minor peaks were unidentified.

28 Quantification ofAnions in Pharmaceuticals Column: IonPac AG14, AS14 Flow Rate: 1.2 mL/min Column: IonPac AG14, AS14 Flow Rate: 1.2 mL/min Eluent: 3.5 mM Sodium carbonate/ Inj. Vol.: 10 µL Eluent: 3.5 mM Sodium carbonate/ Inj. Vol.: 10 µL 0.8 mM Sodium bicarbonate Sample: 10-fold dilution, w/w 0.8 mM Sodium bicarbonate Sample: 100-fold dilution, w/w Detection: Suppressed conductivity, ASRS, Peaks: 1. Unidentified – Detection: Suppressed conductivity, ASRS, Peaks: 1. Unidentified – AutoSuppression recycle mode 2. Unidentified – AutoSuppression recycle mode 2. Unidentified – 3. Chloride 0.08 mg/L 3. Chloride 3.0 mg/L 4 4 4. Bromide 30 2 4. Bromide 1.0 5. Unidentified – 5. Sulfate 0.08 6. Unidentified – 7. Unidentified – 8. Sulfate 0.1 3 µS µS 9. Unidentified –

5 6 7 89 1 2 3 2 5 0 1 4 0

0246810 12 14 16 18 0246810 12 14 Minutes 12745 Minutes 12746

Figure 5. Counterions in cough suppressant. Figure 7. Counterions in multisymptom cold/flu medication.

Column: IonPac AG11, AS11 Sample: 10-fold dilution, w/w Column: IonPac AG11, AS11 Inj. Vol.: 10 µL Eluent: 0.5 mM Sodium hydroxide, Peaks: 1. Unidentified – Eluent: 0.5 mM Sodium hydroxide, hold Sample: 100-fold dilution, w/w hold for 2.5 min; 0.5 mM to 5 mM 2. Unidentified – for 2.5 min; 0.5 mM to 5 mM Peaks: 1. Unidentified – Sodium hydroxide in 3.5 min; 3. Chloride 0.08 mg/L Sodium hydroxide in 3.5 min; 2. Unidentified – 5 mM to 38 mM Sodium 4. Bromide 30 5 mM to 38 mM Sodium hydroxide 3. Chloride 3 mg/L hydroxide in 12 min 5. Benzoate 80 in 12 min 4. Bromide 1 Detection: Suppressed conductivity, ASRS, 6. Unidentified – Detection: Suppressed conductivity, ASRS 5. Unidentified – AutoSuppression recycle mode 7. Unidentified – AutoSuppression recycle mode 6. Sulfate 0.1 Flow Rate: 2 mL/min 8. Sulfate 0.2 Flow Rate: 2 mL/min 7. Citrate 110 Inj. Vol.: 10 µL 9. Phosphate 0.09 8. Saccharin 3 10. Unidentified – 11. Citrate 170 12. Saccharin 120 14 A 18 A 7 11

S 4 µ µS 3 5 45 8 12 1 2 6 1 6 7 8 9 10 2 3 0 0

5 11 B 4 6 12 3 7 1 1 B

9 8

10 µS 4 µS 7 8 5 1 6 2 3 21 0 0

0 2 46810121416 18 0 2 46810121416 18 Minutes Minutes 12747 12748

Figure 6. Anionic ingredients in cough suppressant. Figure 8. Anionic ingredients in multisymptom cold/flu medication.

29 Quantification ofAnions in Pharmaceuticals CONCLUSION REFERENCES Using the IonPac AS11 column with suppressed 1. CFR Title 21, Foods and Drugs, Chapter 1, FDA, conductivity detection, pharmaceutical formulations can B Part 211.22, “Responsibilities of quality control be analyzed for both organic and inorganic anions. In unit.” the same injection, the IonPac AS11 resolves common 2. CFR Title 21, Foods and Drugs, Chapter 1, FDA, I inorganic anions such as fluoride, chloride, bromide, Part 211.160, “General requirements.” sulfate, nitrate, and phosphate, as well as common 3. CFR Title 21, Foods and Drugs, Chapter 1, FDA, I organic anions such as benzoate, sorbate, citrate, and Part 211.165, “Testing and release for distribution.” saccharin. The IonPac AS14 resolves common inorganic 4. Dionex Corporation, “Ion Chromatography in the anions, but is not suitable for common organic anions Pharmaceutical Industry”, Application Note 106. (except acetate). The AS14 column uses an isocratic elu­ 5. Rabin, S.; Stillian, J.R.; Barreto, V.; Friedman, K.; ent and is therefore faster for this analysis. Suppressed and Toofan, M. J. Chromatogr. 1993, 640, 97–109. conductivity eliminates potential interferences from the 6. Dionex Corporation, “Quantification of Carbohy­ nonionic ingredients in the formulation and provides a drates and Glycols in Pharmaceuticals”, Application sensitive means to detect nonchromophoric analytes. Note 117. Inorganic anions can be detected at the 30 to 400-µg/L 7. Dionex Corporation, “Determination of levels, while organic anions can be detected at the 200 Trifluoroacetic Acid (TFA) in Peptides”, to 2000-µg/L levels. Both organic and inorganic anions Application Note 115. were linear over more than two orders of magnitude. The recovery of chloride and bromide from pharmaceu­ LIST OF SUPPLIERS tical formulations was greater than 97% of the labeled Fisher Scientific, 711 Forbes Ave., Pittsburgh, concentrations for the active ingredients. Both methods Pennsylvania, 15219-4785, USA. are also well suited for evaluating trace levels of anionic Tel: 1-800-766-7000. contaminants. Fluka Chemika-BioChemika, Fluka Chemie AG, Indus­ triestrasse 25, CH-9471 Buchs, Switzerland. Tel: 081 755 25 11. Aldrich Chemical Company, Inc., 1001 West Saint Paul Avenue, P.O. Box 355, Milwaukee, Wisconsin, 53233, USA. Tel: 1-800-558-9160. Sigma Chemical Company, P.O. Box 14508, St. Louis, Missouri, 63178, USA. Tel: 1-800-325-3010.

30 Quantification ofAnions in Pharmaceuticals Application Note 123

Determination of Inorganic Anions and Organic Acids in Fermentation Broths

INTRODUCTION mixtures of organic acids and inorganic anions. For com­ Fermentation broths are used in the manufacture plex samples like fermentation broths, the high resolving of biotherapeutics and many other biologically derived power of ion-exchange chromatography and the specific­ products using recombinant genetic technology. Broths ity of suppressed conductivity allow the determination are also used for the production of methanol and ethanol of ionic fermentation broth ingredients, with little interfer­ as alternative energy sources to fossil fuels. In addition, ence from other broth ingredients.3-5 Although biosensor many food and beverage products such as alcoholic and flow-injection analysis methods are commonly used beverages, vinegars, fermented vegetables, sauces, and to evaluate fermentation broths,6, 7 these techniques can­ dairy products are all prepared by controlled fermentation not simultaneously determine multiple compounds. Gel processes. Fermentation monitoring is also important in permeation chromatography with refractive index detec­ detection of spoilage of fruit juices and food products. tion, and anion-exchange chromatography with UV-VIS Recently, attention has been given to characterizing the detection, have been used for analysis of fermentation ingredients of fermentation broths because carbon sources broths, but both are limited by poor selectivity and sensi­ and metabolic by-products can impact the yield of the de­ tivity.8, 9 Anion-exchange chromatography with suppressed sired products. Carbohydrates (glucose, lactose, sucrose, conductivity monitors, by direct injection, a large number maltose, etc.) are carbon sources essential for cell growth of different compounds simultaneously, using a single and product synthesis, while alcohols (ethanol, metha­ instrument and chromatographic method.10 nol, sugar alcohols, etc.), glycols (glycerol), and organic This application note describes the use of two differ­ anions (acetate, lactate, formate, etc.) are metabolic ent anion-exchange columns, with suppressed conductiv­ by-products, many of which reduce desired yields. ity detection, to analyze common organic and inorganic Fermentation broths are complex mixtures of nutrients, anions in yeast and bacterial fermentation broths. The waste products, cells and cell debris, and desired prod­ yeast Saccharomyces cerevisiae in yeast extract-peptone- ucts, such as antibiotics. Many of these ingredients are dextrose (YPD) broth and the bacteria Escherichia coli in nonchromophoric and cannot be detected by absorbance. Luria-Bertani (LB) broth are common fermentation broth Organic and inorganic anions are ionic and therefore cultures and represent eukaryotic and prokaryotic sys­ can be determined by ion chromatography using sup­ tems. Both fermentation broth cultures are complex and pressed conductivity detection. Suppressed conductivity is contain undefined media ingredients, and thus are a great a powerful detection technique with a broad linear range challenge for most separation and detection technologies. and very low detection limits. Nonionic compounds are These formulations also contain carbohydrates, sugar al­ not detected. Suppression lowers the background conduc­ cohols, alcohols, and glycols that have been analyzed us­ tivity caused by the eluent and effectively increases the ing the CarboPac™ PA1, PA10, and MA1 anion-exchange conductivity of the analyte.1, 2 Anion-exchange chroma­ columns with pulsed amperometric detection.11 tography is a technique capable of separating complex

31 Determination of Inorganic Anions and Organic Acids in Fermentation Broths In the methods outlined in this application note, Phenylacetic acid (Sigma Chemical Co.) the selectivities of the IonPac® AS11 and IonPac Propionic acid, sodium salt (Sigma Chemical Co.) AS11-HC anion-exchange columns are compared for Maleic acid, disodium salt (Sigma Chemical Co.) the determination of anionic analytes in fermentation Oxalic acid, sodium salt (Fluka Chemika) broths. The IonPac AS11 column packing consists of l-Malic acid (Eastman Chemical Co.) an alkanol quaternary ammonium latex bonded to a Pyrophosphoric acid (Fluka Chemika) microporous crosslinked ethylvinylbenzene core. The Trichloroacetic acid (Fluka Chemika) AS11-HC (high capacity) latex is bonded to a macro­ Chloroacetic acid (Aldrich Chemical Co.) porous crosslinked ethylvinylbenzene core. Due to the Glycolic acid (Sigma Chemical Co.) greater surface area of its core, the AS11-HC has six l-Glutamic acid (Sigma Chemical Co.) times more anion-exchange capacity than the AS11. Fumaric acid (Fluka Chemika) Both columns are designed for separation of organic d-Gluconic acid, sodium salt (Sigma Chemical Co.) and inorganic anions using sodium hydroxide gradients. Oxalacetic acid (Sigma Chemical Co., and Fluka Organic solvents can be added to eluents to modify the Chemika) selectivity of these columns. Methylmalonic acid (Sigma Chemical Co.) Expected detection limits, linearity, selectivity, 5-Keto-d-Gluconic acid, potassium salt (Sigma stability, and precision for organic and inorganic anions Chemical Co.) in fermentation broths are reported for the IonPac 2-Keto-d-Gluconic acid, hemicalcium salt (Sigma AS11 and AS11-HC columns using the Dionex DX-500 Chemical Co.) BioLC® system with suppressed conductivity detection. Valeric acid (Aldrich Chemical Co.) Isovaleric acid (Sigma Chemical Co.) EQUIPMENT Isobutyric acid (Sigma Chemical Co.) Dionex DX-500 BioLC system consisting of: Sodium bromate (Fluka Chemika) GP40 Gradient Pump with degas option Sodium arsenate, dibasic, 7-hydrate (J.T. Baker ED40 Electrochemical Detector Chemical Co.) LC30 or LC25 Chromatography Oven Sodium acetate, anhydrous (Fluka Chemika) Sodium fluoride (Fisher Scientific) AS3500 Autosampler Sodium nitrate (Fisher Scientific) PeakNet Chromatography Workstation Sodium chloride (Fisher Scientific) Potassium phosphate, dibasic, anhydrous (Fisher REAGENTS AND STANDARDS Scientific) Reagents Citric acid, monohydrate (Fisher Scientific) Sodium hydroxide, 50% (w/w) (Fisher Scientific and Sodium bromide (Aldrich Chemical Co.) J. T. Baker) Sodium sulfate, anhydrous (EM Science) Deionized water, 18 MΩ-cm resistance or higher was Sodium carbonate, monohydrate (Fisher Scientific) used for preparing all standards and eluents. Water that was used to prepare YPD broth was filter ster­ Culture and Media ilized by passage through a 0.2-µm filter. Bacto YPD Broth (DIFCO Laboratories, Cat# 0428-17-5) Standards Bacto Yeast Extract (DIFCO Laboratories, Lactic acid (Fisher Scientific) Cat# 0127-15-1) Succinic acid (Aldrich Chemical Co.) Bacto Peptone (DIFCO Laboratories, Cat# 0118-15-2) Pyruvic acid, sodium salt (Fisher Scientific) LB Broth (DIFCO Laboratories, Cat# 0446-17-3) dl-Isocitric acid, trisodium salt (Sigma Chemical Co.) Yeast, S. cerevisiae; Bakers Yeast type II (Sigma n-Butyric acid, sodium salt (Sigma Chemical Co.) Chemical Co., Cat# 45C-2) Sodium formate (Fisher Scientific) Bacteria, E. coli (donated by SRI International)

32 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Table 1. Chromatographic Conditions

Conditions System 1 System 2 Column: IonPac AS11 Analytical (P/N 44076) IonPac AS11-HC Analytical (P/N 52960) IonPac AG11 Guard (P/N 44078) IonPac AG11-HC Guard (P/N 52962) ATC-1 Anion Trap Column (P/N 37151) ATC-1 Anion Trap Column (P/N 37151)

Flow Rate: 2.0 mL/min 1.5 mL/min Injection Volume: 10 µL 10 µL Oven Temperature: Ambient 30 °C Detection (ED40): Suppressed conductivity, ASRS®, Suppressed conductivity, ASRS, AutoSuppression® recycle mode, AutoSuppression recycle mode, 300 mA 300 mA Eluents: A: Water A: Water B: 5 mM sodium hydroxide B: 5 mM sodium hydroxide C: 100 mM sodium hydroxide C: 100 mM sodium hydroxide Gradient: 0.5–38 mM sodium hydroxide: 1–60 mM sodium hydroxide: 0.5 mM sodium hydroxide, hold for 2.5 min; 1 mM sodium hydroxide, hold for 8 min; 0.5–5 mM sodium hydroxide in 3.5 min; 1–15 mM sodium hydroxide in 10 min; 5–38 mM sodium hydroxide in 12 min. 15–30 mM sodium hydroxide in 10 min. 30–60 mM sodium hydroxide in 10 min; 60 mM sodium hydroxide, hold for 2 min.

Method: Time Time (min) A (%) B (%) C (%) (min) A (%) B (%) C (%) 0.0 90 10 0 0.0 80 20 0 2.5 90 10 0 8.0 80 20 0 6.0 0 100 0 18.0 85 0 15 18.0 0 62 38 28.0 70 0 30 18.1 90 10 0 38.0 40 0 60 25.0 90 10 0 40.0 40 0 60 40.1 80 20 0 50.0 80 20 0 Typical Background 0.5 mM sodium hydroxide: 0.5–1 µS 1 mM sodium hydroxide: 0.5–1 µS Conductivity: 38 mM sodium hydroxide: 2–3 µS 60 mM sodium hydroxide: 2–3 µS Typical System Operating Backpressure: 12.4 Mpa (1800 psi) 15.2 Mpa (2200 psi)

CONDITIONS Carbonate can be removed by placing an anion trap See “Conditions” (Table 1). column (ATC-1, P/N 37151) between the pump and the injection valve. Commercially available sodium PREPARATION OF SOLUTIONS AND REAGENTS hydroxide pellets are covered with a thin layer of so­ Sodium Hydroxide Eluents dium carbonate and should not be used. A 50% (w/w) 5 mM Sodium Hydroxide sodium hydroxide solution is much lower in carbonate It is essential to use deionized water of high resis­ and is the preferred source for sodium hydroxide. tance (18 MΩ-cm) that is as free of dissolved carbon Dilute 0.524 mL of 50% (w/w) sodium hydroxide dioxide as possible. Carbonate is formed in alkaline el­ solution into 2000 mL of thoroughly degassed water to uents from carbon dioxide. Carbonate, a divalent anion yield 5 mM sodium hydroxide. Keep the eluents blan­ at high pH, binds strongly to the columns and causes a keted under 5–8 psi (34–55 kPa) of helium at all times. loss of chromatographic resolution and efficiency.

33 Determination of Inorganic Anions and Organic Acids in Fermentation Broths 100 mM Sodium Hydroxide Heat-inactivated yeast fermentation broth super­ Follow the same precautions described above for natant was spiked with anions for the recovery and the 5 mM sodium hydroxide eluent. Dilute 10.4 mL of stability study. To inactivate the culture, broth super­ 50% (w/w) sodium hydroxide solution into 1990 mL natant was diluted 10-fold, and heated in boiling water of thoroughly degassed water to yield 100 mM sodium for 10 min. An aliquot of heat-inactivated supernatant hydroxide. Keep the eluents blanketed under 5–8 psi was then diluted another 10-fold using 100 µg/mL (34–55 kPa) of helium at all times. lactate, acetate, formate, pyruvate, sulfate, oxalate, phosphate, and citrate. The final concentration of each Stock Standards anion was 10 µg/mL. Another aliquot of heat-inacti­ Solid standards were dissolved in water to 10 g/L vated yeast culture supernatant was diluted 100-fold in anionic concentrations. These were combined and water, serving as an unspiked “blank”. further diluted with water to yield the desired stock mixture concentrations. E. Coli Fermentation Broth Culture—Standard Media The solutions were kept frozen at –20 °C until needed. LB Broth is dissolved to a concentration of 25 g/L For determinations of linear range, combine 10-g/L with water, heated to a boil, and autoclaved for solutions of chloride, bromide, and citrate to make a 15 minutes at 121 psi. A liter of LB broth contains 1-mg/L standard mix solution. Dilute with water to 10 g of tryptone, 5 g yeast extract, and 10 g of sodium concentrations of 800, 600, 400, 200, 100, 80, 60, 40, chloride per 25 g. The culture was incubated and 20, 10, 4, and 1 µg/L. Standard solutions of acetate, sampled as described for the yeast standard media. bromide, nitrate, sulfate, phosphate, and citrate were also prepared for estimating lower detection limits and RESULTS AND DISCUSSION linearity at concentrations of 10, 8, 6, 4, 2, 1, 0.8, 0.6, Selectivity 0.4, 0.2, 0.1, 0.08, 0.06, 0.04, 0.02, and 0.01 mg/L. IonPac AS11 Chloride was prepared at concentrations of 2.4, 1.9, Figure 1A shows the separation of the common 1.5, 1.0, 0.49, 0.24, 0.19, 0.15, 0.098, 0.048, 0.024, fermentation broth anions using an IonPac AS11 0.019, 0.014, 0.0097, 0.0048, and 0.0024 mg/L. column set with a 0.5–38 mM NaOH gradient (Table 1, System 1) flowing at 2.0 mL/min. The organic and SAMPLE PREPARATION inorganic anions were well-resolved. The analytes Yeast Fermentation Broth Culture—Standard Media were eluted from the column in less than 20 min. The In a sterile 500-mL Erlenmeyer flask, dissolve retention times of the anions in Figure 1A are listed 10 g of Bacto YPD Broth (DIFCO Laboratories, Cat# in Table 2. In general, monovalent anions eluted first, 0428-17-5) in 200 mL filter-sterilized water. Bacto followed by di- and trivalent anions. YPD Broth contains 2 g Bacto Yeast Extract, 4 g Bacto Peptone, and 4 g dextrose (glucose) per 10 g. Dissolve IonPac AS11-HC 1.0 g yeast (S. cerevisiae; Bakers Yeast type II; Sigma Figure 1B shows the analysis of common fermenta­ Chemical Co., Cat# 45C-2) in the YPD broth. Cap tion broth anions using the IonPac AS11-HC column. the flask with a vented rubber stopper. Incubate the Analytes were eluted using a 1–60 mM sodium hydrox­ culture in a 37 °C shaking water bath (500–600 rpm) ide gradient (Table 1, System 2) flowing at 1.5 mL/min. for 24 h, removing aliquots at designated time points A stronger eluent was needed to elute anions from this and placing them on ice. For this study, samples were column due to its higher capacity. The higher capacity taken after the addition of yeast at 0, 0.5, 1, 2, 3, 4, 5, improves resolution of early eluting peaks. For example, 6, 7, and 24-h intervals. The incubation starts when lactate and acetate are better resolved on the AS11-HC yeast is added to the medium. Aliquots are centrifuged than the AS11. The elution order of the AS11-HC is at 14,000 × g for 10 min and diluted 10- and 100-fold similar to the AS11. Table 2 also summarizes the reten­ in purified water. Diluted supernatant (10 µL) was tion times of different anions on the AS11-HC column. analyzed directly. These results demonstrate that the AS11-HC column has

34 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Column: IonPac AS11, AG11 Peaks: 1. Lactate Eluent: 0.5 mM sodium hydroxide, 2. Acetate Table 2. Retention Times for Common hold for 2.5 min; 3. Propionate Organic and Inorganic Anions 0.5–5 mM sodium hydroxide 4. Formate in 3.5 min; 5. 2-Keto-d-Gluconate 5–38 mM sodium hydroxide 6. Pyruvate in 12 min. 7. Valerate Retention Times (Minutes) 8. Monochloroacetate Flow Rate: 2.0 mL/min Analyte IonPac IonPac 9. Bromate Inj. Volume: 10 µL 10. Chloride AS11/AG11 AS11-HC/AG11-HC Temperature: Ambient 11. Phenylacetate Fluoride 2.3 8.7 Sample: 10 mg/L each analyte 12. Bromide 13. Nitrate Gluconate 2.3 8.2 14. Malate 15. Methylmalonate Lactate 2.5 8.8 16. Carbonate Acetate 2.6 9.5 17. Malonate 14 A 18. Maleate Glycolate 2.6 9.4 19. Sulfate Propionate 2.9 11.0 20. Oxalate 10 21. Trichloroacetate Isobutyrate 3.2 12.3 22. Phosphate Formate 3.4 12.4 23. Citrate Butyrate 3.6 12.8 µS 19 24. Isocitrate 25. Pyrophosphate 13 16,17 20 2-Keto-d-Gluconate 4.0 13.1 14 Pyruvate 4.3 13.5 12 15 22 2 4 18 24 1 8 23 Isovalerate 4.3 13.7 6 9 11 25 3 21 5 7 Valerate 5.1 14.8 0 Monochloroacetate 5.5 15.3 0 5110 5 20 Bromate 5.8 16.1 Minutes Chloride 6.1 16.7 Phenylacetate 8.0 21.7 Column: IonPac AS11-HC, AG11-HC Bromide 8.2 21.9 Eluent: 1 mM sodium hydroxide, hold for 8 min; 1–15 mM sodium hydroxide in 10 min; 5-Keto-d-Gluconate 8.3 20.1 15–30 mM sodium hydroxide in 10 min; Nitrate 8.4 22.4 30–60 mM sodium hydroxide in 10 min; 60 mM sodium hydroxide, hold for 2 min. Glutarate N/A 22.5 Flow Rate: 1.5 mL/min Succinate 10.1 22.9 Inj. Volume: 10 µL Malate 10.1 23.0 Temperature: 30 ºC 8 B Sample: 10 mg/L each analyte Carbonate N/A 23.5 Methylmalonate 10.2 23.4 10 Malonate 10.4 23.8 Maleate 10.7 24.9 µS Sulfate 11.0 25.4 14 19 13 15,16 20 Oxalate 11.4 26.6 17 22 24 11,12 18 23 21 Fumarate 11.4 26.8 4 8 25 1 6 9 Oxalacetate 12.2 29.2 2 5 3 7 Trichloroacetate 13.5 39.0 0 Phosphate 13.9 31.8 0 51510 20 25 30 35 40 45 Arsenate 15.1 33.9 Minutes Citrate 15.5 34.4 13618 Isocitrate 16.0 35.3 Figure 1. Common organic and inorganic anions found in Pyrophosphate 19.7 39.1 fermentation broths analyzed on the IonPac AS11 and AS11-HC columns with suppressed conductivity. N/A - Not available

slightly different selectivity than the AS11. For example, trichloroacetate elutes before phosphate on the AS11, but the AS11 column elutes phenylacetate, bromide, and elutes after pyrophosphate on the AS11-HC. The high nitrate several minutes before malate; the AS11-HC capacity of the AS11-HC permits larger sample loads. elutes these compounds much closer to malate. Also,

35 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Figure 2A shows early-eluting peaks from the Table 3. Estimated Lower Detection Limits analysis of 24-µg sample of fermentation broth anions (10-µL Injection) analyzed on the AS11, and Figure 2B shows the same System 1 AS11 analysis on the AS11-HC. At this sample load, the ng µg/L AS11 is overloaded. Acetate 2 200 Chloride 0.5 50 Detection Limits Bromide 4 400 Nitrate 3 300 The detection limits for a 10-µL injection of rep­ Sulfate 1 100 resentative fermentation broth anions, in the absence Phosphate 4 400 of broth matrix, using the AS11 column, are shown in Citrate 4 400 Table 3. The detection limit is defined as the minimum concentration required to produce a peak height signal- to-noise ratio of 3. The detection limit can be further decreased by increasing the injection volume above the Column: IonPac AS11, AG11 Eluent: 0.5 mM sodium hydroxide, hold for 2.5 min; 10-µL injection volume used for this application note. If 0.5–5 mM sodium hydroxide in 3.5 min; increasing injection volume also increases sample load 5–38 mM sodium hydroxide in 12 min. Flow Rate: 2.0 mL/min beyond the AS11 column capacity, the higher capacity Inj. Volume: 10 µL AS11-HC can overcome this limitation. The detec­ Temperature: Ambient tion limit can be further decreased by using smoothing Sample: 100 mg/L each analyte (24 µg total load) Peaks: 1. Lactate algorithms available in PeakNet software and by using 2. Acetate 70 external water mode. A 3. Propionate 10 4. Formate 5. 2-Keto-d-Gluconate 6. Pyruvate 7. Valerate 8. Monochloroacetate Table 4. Peak Area Precision (RSD, %) µS 9. Bromate 10. Chloride Analyte Last First Second 96 Hour 8 9 8 Hours 48 Hours 48 Hours Period 3,4 1,2 5 6 7 Lactate 0.2 0.4 0.3 0.5 0 Acetate 0.2 0.7 0.4 0.6 Formate 0.1 0.3 0.2 0.4 0 2325476 Minutes Pyruvate 0.5 0.5 0.7 0.8 Chloride 0.6 0.5 0.4 0.5 Column: IonPac AS11-HC, AG11-HC Sulfate 0.3 1.2 1.3 1.3 Eluent: 1 mM sodium hydroxide, hold for 8 min; Oxalate 0.4 0.8 0.5 1.2 1–15 mM sodium hydroxide in 10 min; 15–30 mM sodium hydroxide in 10 min. Phosphate 0.7 1.6 0.5 1.9 70 30–60 mM sodium hydroxide in 10 min; Citrate 0.3 1.8 0.6 2.1 B 60 mM sodium hydroxide, hold for 2 min. Flow Rate: 1.5 mL/min 10 Retention Time Precision (RSD, %) Inj. Volume: 10 µL Temperature: 30 ºC Analyte Last First Second 96 Hour Sample: 100 mg/L each analyte 8 Hours 48 Hours 48 Hours Period µS (24 µg total load) Lactate 0.1 0.2 0.2 0.3 4 8 Acetate 0.1 0.2 0.1 0.3 6 9 Formate 0.1 0.2 0.1 0.3 1 2 5 3 7 Pyruvate 0.0 0.2 0.1 0.3 0 Chloride 0.0 0.2 0.1 0.4 6 8110 2 14 16 18 Sulfate 0.0 0.3 0.1 0.7 Minutes Oxalate 0.0 0.3 0.1 0.7 13619 Phosphate 0.0 0.3 0.1 0.7 Citrate 0.0 0.3 0.0 0.7 Figure 2. Separation of early eluting organic and inorganic anions at high levels (24 µg total load) using the IonPac AS11 and AS11-HC.

36 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Acetate; r2 = 0.9991 10-µL Injection 500000 Chloride; r2 = 0.9999

2 400000 Bromide; r = 0.9998

Nitrate; r2 = 0.9997 300000 2 Area Units Sulfate; r = 0.9998 200000 Phosphate; r2 = 0.9999 100000 Citrate; r2 = 0.9998 0

0 20 40 60 80 100 120 140 ng Injected 14255

Figure 3. Method linearity for IonPac AS11 with suppressed conductivity detection.

500000

450000

400000

350000 Lactate; RSD = 0.5% 300000 Acetate; RSD = 0.6% 250000 Formate; RSD = 0.4%

Area Units 200000 Pyruvate; RSD = 0.8% Chloride; RSD = 0.5% 150000 Sulfate; RSD = 1.3% 100000 Oxalate; RSD = 1.2% 50000 Phosphate; RSD = 1.9% Citrate; RSD = 2.1% 0 0 20 40 60 80 96 Time (h) 14256

Figure 4. Peak Areas during 4 day repetitive analysis of heat-inactivated yeast fermentation broth.

Linearity Table 5. Recovery of Anions in the Yeast Chloride, bromide, and citrate standards ranging Fermentation Broth from 1–1000 mg/L (10–10,000 ng) were injected (in Analyte Percent Recovery triplicate) on the AS11 column. For these analytes, the Lactate 100 peak area response was found to be linear over this Acetate 88 range (r2 ≥ 0.999). Acetate, nitrate, sulfate, and phos­ Formate 101 phate were investigated over the concentration range of Pyruvate 99 0.1–12 mg/L (1–120 ng) and showed high linearity Sulfate 101 2 Phosphate 100 (r ≥ 0.999). Broad linear ranges help reduce the need Citrate 84 to repeat sample analyses when components vary greatly in concentration. Representative calibration curves for the AS11 column is presented in Figure 3.

37 Determination of Inorganic Anions and Organic Acids in Fermentation Broths 40

35 Lactate; RSD = 0.3% Acetate; RSD = 0.3% 30 Formate; RSD = 0.3% 25 Pyruvate; RSD = 0.3% Chloride; RSD = 0.4% 20 Carbonate; RSD = 0.7% 15 Sulfate; RSD = 0.7% Retention Time (Min) Oxalate; RSD = 0.7% 10 Phosphate; RSD = 0.7% 5 Citrate; RSD = 0.7%

0 0 20 40 60 80 96 Time (h) 14257

Figure 5. Retention times during 4 day repetitive analysis of heat-inactivated yeast fermentation broth.

Precision and Stability Column: IonPac AS11, AG11 Peaks: 1. Unknown The peak area and retention time RSDs were Eluent: 0.5 mM sodium hydroxide, 2. Lactate hold for 2.5 min; 3. Acetate determined for replicate injections of common anions 0.5–5 mM sodium hydroxide 4. Unknown in 3.5 min; 5–38 mM 5. Formate spiked into yeast fermentation broth. Anion standards Sodium hydroxide in 12 min. 6. Unknown 7. Unknown were added to heat-inactivated S. cerevisiae fermentation Flow Rate: 2.0 mL/min 8. Chloride Inj. Volume: 10 µL 9. Unknown broth culture supernatant to yield 10 mg/L spike concen­ Temperature: Ambient 10. Unknown trations and then analyzed repeatedly for 96 h (10-µL Detection: Suppressed conductivity, ASRS 11. Malate/Succinate AutoSuppression recycle mode 12. Malonate/Carbonate injections) on the AS11-HC column. Statistics for this 13. Sulfate Sample: S. cerevisiae culture supernatant, 14. Fumarate/Oxalate experiment are presented in Table 4. Figures 4 and 5 Diluted 100-fold 15. Unknown 3 16. Phosphate show peak areas and retention times for every injec­ A 17. Maleate 8 18. Unknown tion of this experiment. Peak area RSDs were 0.4–2.1% over 96 h. Retention time RSDs ranged from 0.3–0.7%. 16 Retention times shifted slightly at 45 h (Figure 5) when µS 13 the 100 mM sodium hydroxide eluent was replenished 12 11 14 during the study. These results demonstrate that chang­ 2 9 10 1 5 ing eluents can affect retention time precision. 3 4 6 7 0 Recovery from Sample Matrix 0 5110 5 20 After correction for endogenous amounts, the mea­ Minutes 3 B sured levels of selected anions spiked into a heat-inacti­ 8 vated yeast fermentation broth culture were compared to their expected levels. These results are presented in Table 5, and show good recovery of anions from the yeast µS fermentation broth. 3 12 18 2 14 17 11 1 10 13 15 16 5 9 Yeast (S. cerevisiae) Culture

0 Yeast were grown in Bacto YPD broth at 37 °C for up to 24 h. Figure 6 shows the separation of fermenta­ 0 5110 5 20 Minutes 13621 tion broth ingredients in a yeast culture at the begin­ ning (Figure 6A) and after 24 h (Figure 6B) of incu­ Figure 6. S. cerevisiae fermentation broth culture (100-fold dilu- bation. Lactate, acetate/glycolate, formate, valerate, tion) using the IonPac AS11 column at 0 h (A) and 24 h (B) of incubation. methylmalonate, and citrate increased during the 24-h

38 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Table 6. Anions in Yeast Fermentation Broth During in 24 h Incubation

Broth Concentration (µg/mL)

Incubation Time (h) 0 0.5 1 2 3 4 5 6 7 24 Lactate 59 67 66 70 88 85 84 89 90 338 Acetate 72 122 153 187 199 222 227 247 235 704 Propionate 11 10 4 9 4 6 4 7 4 11 Formate 7 10 11 13 11 14 6 7 6 21

2-Keto-d-Gluconate 4 9 2 10 0 0 0 11 4 2 Pyruvate 10 14 19 24 14 16 17 17 17 6 Valerate 0 0 0 0 0 5 5 4 11 24 Anions Chloride 348 345 353 320 355 356 357 354 371 347 Malate 0 7 13 9 8 15 15 12 15 11 Methylmalonate 100 125 169 180 224 248 247 237 253 229 Malonate 428 452 569 476 474 547 563 531 577 563 Sulfate 68 68 79 63 67 65 64 63 61 57 Oxalate 12 14 14 10 12 13 12 16 15 11 Phosphate 165 124 92 62 55 57 58 58 59 62 Citrate 0 0 0 13 15 12 13 13 10 0

incubation. Table 6 lists the measured concentrations Bacteria (E. coli) Culture of these and other analytes during the 24-h incubation. Bacteria (E. coli) was grown on LB broth for Between 7 and 24-h, no additional time points were 24 h at 37 ºC. Figure 8A shows the anions present taken; however, substantial increases in the levels of in this broth at the beginning of the culture, and lactate, acetate/glycolate, formate, and valerate occur. Figure 8B shows anions after 24 h. Some anions remained constant throughout the 24-h To examine anions at lower concentrations, injec­ incubation, including chloride, malonate, sulfate, and tions of a more concentrated culture are needed and oxalate/fumarate. Phosphate concentration decreased, the AS11-HC column is the best choice. Yeast fermen­ presumably due to incorporation into the biomass tation broth (diluted only 10-fold) was analyzed by (e.g., DNA, RNA, membrane phospholipids, etc.). At both the AS11 and AS11-HC columns, and is present­ least 10 unidentified peaks were observed. The area ed in Figures 7A and 7B, respectively. units for eight of these peaks changed over the course The AS11 column did not resolve the first unknown of the incubation period. Changes in lactate, acetate, peak from lactate, while the AS11-HC did. Lactate and formate concentrations are expected as a result of and acetate were better resolved on the AS11-HC. normal metabolic processes. Trending can be used to Butyrate was resolved from formate on the track culture status.

39 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Column: IonPac AS11, AG11 Peaks: 1. Unknown Column: IonPac AS11, AG11 Peaks: 1. Unknown Eluent: 0.5 mM sodium hydroxide, hold for 2.5 2. Lactate Eluent: 0.5 mM sodium hydroxide, 2. Unknown min; 0.5–5 mM sodium hydroxide in 3.5 min; 3. Acetate/Glycolate hold for 2.5 min; 3. Unknown 4. Formate 4. Lactate 5–38 mM sodium hydroxide in 12 min. 0.5–5mM sodium hydroxide 5. Butyrate 5. Acetate Flow Rate: 2.0 mL/min in 3.5 min; 6. Pyruvate/Isovalerate 6. Propionate Inj. Volume: 10 µL 7. Valerate 5–38 mM sodium hydroxide 7. Formate Detection: Suppressed conductivity, ASRS 8. Chloride in 12 min. 8. Valerate AutoSuppression recycle mode Flow Rate: 2.0 mL/min 9. Chloride Temperature: Ambient Inj. Volume: 10 µL 10. Unknown 11. Phenylacetate Detection: Suppressed conductivity, 10 12. Bromide A ASRS, AutoSuppression 13. Nitrate/5-Keto-d-Gluconate recycle mode 8 14. Unknown Sample: E. coli culture supernatant, 15. Unknown 4 diluted 10-fold 16. Malate/Succinate 2 17. Malonate/Carbonate µS 3 18. Sulfate 1 5 19. Fumarate/Oxalate 20. Phosphate 2 A 9 6 7 20 0

µS 0 1325476 3 4 Minutes 17 14 19 16 18 Column: IonPac AS11-HC, AG11-HC 2 12 15 5 11 13 Eluent: 1 mM sodium hydroxide, hold for 8 min; 1 7 1–15 mM sodium hydroxide in 10 min; 0 15–30 mM sodium hydroxide in 10 min; 30–60 mM sodium hydroxide in 10 min; 60 mM sodium hydroxide, hold for 2 min. 0 5110 5 20 Flow Rate: 1.5 mL/min Minutes Inj. Volume: 10 µL 20 2 Detection: Suppressed conductivity, ASRS AutoSuppression B 9 recycle mode Temperature: 30 °C 5 16

10 B µS 8 3 17 19 18 4 13 15 1112 14 6 1 2 8 10 µS 0 3 2 4 1 0 5110 5 20 5 Minutes 6 7 13623 0 Figure 8. E. coli fermentation broth culture using the IonPac AS11 column at 0 h (A) of incubation and the AS11 column at 6 8110 2 14 16 18 Minutes 24 h (B) of incubation. 13622

Figure 7. S. cerevisiae fermentation broth culture (10-fold dilution) using the IonPac AS11 and AS11-HC column at 24 h sample using the AS11-HC column. Concentrations of of incubation. anions were determined at different time points during incubation. After 24 h, lactate decreased, while acetate increased in concentration. Malate/succinate increased AS11-HC column. Furthermore, many of the trace over this period. Chloride remained unchanged. Peaks components that could not be measured using a 100- having retention times equal to propionate and valerate fold dilution could be measured with a 10-fold diluted were present after 24 h of incubation.

40 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Conclusion 7. Huang, Y. L.; Khoo, S. B.; Yap, M. G. S. “Flow- These results show that both yeast and bacterial injection analysis-wall-jet electrode system for culture fermentation broths can be analyzed for monitoring glucose and lactate in fermentation anion composition using ion chromatography and broths”. Anal. Chim. Acta 1993, 283, 763–771. suppressed conductivity. Two columns (IonPac AS11 8. Schugerl, K.; Brandes, L.; Xiaoan, W.; Bode, J.; and AS11-HC) are available for fermentation broth Ree, J. I.; Brandt, J.; Hitzmann, B. “Monitoring and analysis of organic acids and inorganic anions. The control of recombinant protein production”. AS11-HC permits higher sample loading due to higher Anal. Chim. Acta 1993, 279, 3–16. capacity. The high capacity of the column is able to 9. Gey, M.; Nagel, E.; Weissbrodt, E.; Stottmeister, U. resolve lactate, acetate, and formate. Complex mix­ “Fast liquid chromatographic determination of or­ tures of organic and inorganic anions can be monitored ganic acids in fermentation media with short glass simultaneously during fermentation, providing the columns”. Anal. Chim. Acta 1988, 213, 227–230. analyst with some of the information needed to opti­ 10. Buttler, T.; Gorton, L.; Marko-Varga, G. “Charac­ mize the fermentation. terization of a sampling unit based on tangential flow filtration for -on line bioprocess monitoring”. REFERENCES Anal. Chim. Acta 1993, 279, 27–37. 1. Dionex Corporation, “Ion Chromatography in the 11. Dionex Corporation, “Determination of carbo­ Pharmaceutical Industry”. Application Note 106; hydrates, alcohols, and glycols in fermentation Sunnyvale, CA broths”. Application Note 122; Sunnyvale, CA 2. Rabin, S.; Stillian, J.R.; Barreto, V.; Friedman, K.; LIST OF SUPPLIERS and Toofan, M.J. “New membrane-based electro­ J. T. Baker Incorporated, 222 Red School Lane, Phil­ lytic suppressor device for suppressed conductivity lipsburg, NJ 08865 USA, Tel: 1-800-582-2537, detection in ion chromatography” J. Chromatogr. www.jtbaker.com 1993, 640, 97–109. Eastman Chemical Company, 1001 Lee Road, Roches­ 3. Robinett, R. S. R.; George, H. A.; Herber, W.K., ter, NY, 14652-3512 USA, Tel: 1-800-225-5352, “Determination of inorganic cations in fermentation www.eastman.com and cell culture media using cation-exchange liquid EM Science, P.O. Box 70, 480 Democrat Road, chromatography and conductivity detection”. Gibbstown, NJ, 08027 USA, Tel: 1-800-222-0342, J. Chromatogr., A 1995, 718, 319–327. www.emscience.com 4. Joergensen, L.; Weimann, A.; Botte, H.F. “Ion chro­ Fisher Scientific, 711 Forbes Avenue, Pittsburgh, PA matography as a tool for optimization and control 15219-4785 USA, Tel: 1-800-766-7000, www. of fermentation processes”. J. Chromatogr., 1992, fischersci.com 602, 179-188. Fluka Chemika, Fluka Chemie AG, Industriestrasse 5. Loconto, P.R.; Hussain, N. “Automated coupled ion 25, CH-9471 Buchs, Switzerland, Tel: 081 755 25 exclusion-ion chromatography for the determina­ 11, www.sigmaaldrich.com tion of trace anions in fermentation broth”. Sigma-Aldrich Chemical Company, P.O. Box 14508, J. Chromatogr. Sci. 1995, 33, 75–81. St. Louis, MO 63178 USA, Tel.: 1-800-325-3010, 6. Forman, L. W.; Thomas, B. D.; Jacobson, F. S. www.sigmaaldrich.com “On-line monitoring and control of fermentation processes by flow-injection analysis”. Anal. Chim. Acta 1991, 249, 101–111.

41 Determination of Inorganic Anions and Organic Acids in Fermentation Broths Application Note 160

Determination of Residual Trifluoroacetate inDetermination Protein Purification of Residual BuffersTrifluoro andacetate in Protein PeptidePurification Preparations Buffers and by Reagent-FreePeptide Preparations by ™ IonReagent-Free Chromatography Ion Chromatography

INTRODUCTION EQUIPMENT A Reagent-Free ion chromatography (RFIC) system Dionex Ion Chromatography system (ICS-2000 or allows the determination of anions or cations with only ICS-2500) consisting of: the addition of deionized water. For anion analysis, the GP50 Gradient Pump RFIC system prepares high-purity, carbonate-free CD25A Conductivity Detector potassium hydroxide eluents. After separation on the EG50 Eluent Generator with EluGen® EGC II anion-exchange column, an anion self-regenerating KOH cartridge (P/N 060585) suppressor automatically suppresses the eluent and the AS50 Autosampler with thermal compartment sample anions are detected by suppressed conductivity. RFIC allows rapid method development and easy Columns: IonPac AS18 analytical, 4 × 250 mm transfer of methods to other labs. In this application (P/N 060549) note, we used RFIC to determine the concentration of IonPac AG18 guard, 4 × 50 mm residual trifluoroacetate (TFA) in samples of interest to (P/N 060551) the pharmaceutical and biotechnology industries. 50-µL sample loop (P/N 42950) or 100-µL sample loop TFA is commonly used during the purification of (P/N 42951) pharmaceutical and biotechnology products. For Suppressor: ASRS® ULTRA II, 4 mm (P/N 61561) example, TFA is used with an acetonitrile gradient on a CR-ATC (Continuously Regenerated Anion Trap Column) preparative reversed-phase HPLC column to purify (P/N 060477) synthetic peptides. Because TFA is toxic, its removal Chromeleon® Chromatography Workstation (Release 6.5 must be reliably measured in products intended for and higher) preclinical or clinical applications. The high-capacity IonPac® AS18 anion-exchange column was used to separate trace TFA from an excess of chloride, phos- REAGENTS AND STANDARDS phate, and other anions in three different pharmaceutical Deionized water, Type I reagent-grade, 18 MΩ-cm buffers. The method presented in this application note resistance expands on the work presented in Application Note 115, Sodium trifluoroacetate (trifluoroacetic acid, sodium salt) “Determination of Trifluoroacetate (TFA) in Peptides”, (Aldrich P/N 13,210-1) that described the use of a carbonate/bicarbonate eluent with the IonPac AS14. This new method, based on the use of RFIC with an IonPac AS18 column, improves the sensitivity of TFA determinations and allows more samples to be analyzed directly.

42 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography SAMPLES SEPARATION METHOD Phosphate-Buffered Saline, pH 7.4 (Invitrogen Life Time EG50 SRS Technologies catalog number 10010) (min) Conc Current • 1.06 mM potassium phosphate, monobasic (mM) (mA)

(KH2PO4) 0.00 22.0 80 • 155.17 mM sodium chloride Comments: Load sample loop, 80 mA SRS current • 2.96 mM sodium phosphate, dibasic 7-hydrate setting, Acquisition ON

(Na2HPO4•7H2O) 6.00 22.0 80 Protein Purification Buffer #14 6.01 28.0 80 • 0.1 M acetic acid, pH 3 Comment: Step to 28 mM KOH • 0.25 M sodium chloride 12.00 28.0 80 • 0.01% Tween® 20 •1 mg/mL bovine serum albumin (Fluka P/N 05468) 12.01 50.0 80 Comment: Step to 50 mM KOH for cleanup Protein Purification Buffer #24 • 0.1 M Tris, pH 7.4 14.00 50.0 124 Comment: Step to 124 mA SRS current • 0.14 M sodium chloride • 0.01 % Tween 20 15.00 50.0 124 •1 mg/mL bovine serum albumin (Fluka P/N 05468) 15.01 22.0 124 Commercial Peptide – Human Angiotensin II Comment: Step back to 22 mM KOH (Sigma-Aldrich P/N A9525) 17.00 22.0 80 •Asp – Arg – Val – Tyr – Ile – His – Pro – Phe, Comment: Step back to 80 mA SRS current Acetate salt 20.00 22.0 80 Comment: Acquisition OFF CONDITIONS Eluent: Potassium hydroxide (EG50 as the PREPARATION OF SOLUTIONS AND REAGENTS source) Temperature: 30 °C TFA Stock Standard Solution 1000 µg/mL Dissolve 0.1203 g of sodium trifluoroacetate in Eluent Flow Rate: 1.0 mL/min deionized water and dilute to 100 mL in a volumetric Detection: Suppressed conductivity, ASRS flask. Dilute this stock standard solution to the desired ULTRA II, recycle mode concentrations. ASRS Current Setting: See method SYSTEM PREPARATION AND SETUP Expected Background This section describes the procedures for the initial Conductivity: <1 µS (22 mM KOH) installation and start-up of the ASRS ULTRA II, EGC II Typical System KOH EluGen cartridge, and CR-ATC. Prepare the Backpressure: 14 MPa (2000 psi) to ASRS according to the Quickstart Instructions for the 17.2 MPa (2500 psi) ASRS ULTRA II (Document No. 031951). Install the EGC II OH EluGen cartridge according to the instruc- Sample Volume: 5–100 µL tions in the Operator’s Manual for the EG50 Eluent Generator System (Document Number 031908). Install the CR-ATC between the EGC II KOH cartridge and the degas module in the EG50 according to the Operator’s Manual for the Continuously Regenerated Anion Trap Column (Document Number 031910).

43 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography Connect the columns and suppressor in the IC system by using the black PEEK 0.010-in. (0.25-mm) Column:IonPac AG18 and AS18 Eluent: 22 mM KOH for 0–6 min, tubing. Keep the lengths of connecting PEEK tubing as 28 mM KOH for 6–12 min, short as possible to minimize the system void volume 50 mM KOH for 12–15 min, 22 mM KOH for 15–20 min and thus ensure efficient chromatographic performance. Eluent Source: EG50 Carefully use a plastic tubing cutter to ensure the tubing Eluent Flow Rate:1.0 mL/min cuts have straight, smooth surfaces. Irregularity on the Oven Temperature: 30 °C Suppressor: ASRS, AutoSuppression, recycle mode surface of a tubing end can result in unwanted addi- ASRS ULTRA Current: 80 mA for 0–14 min, tional dead volume. 124 mA for 14.01–17 min, 80 mA for 17.01– 20 min Inj. Volume:5 µL SYSTEM OPERATION Peaks: 1. Fluoride 2 mg/L Turn on the gradient pump to begin the flow of 5 2. Chloride 4 eluent through the system. If the system backpressure is 2 3. Nitrite 10 4. Carbonate – below 14 MPa (2000 psi), a length of yellow PEEK 5. Bromide 10 9 0.003-in. (0.075-mm) tubing should be added between 6. Sulfate 10 the outlet of the degas assembly in the EG50 and the 7. Nitrate10 3 8. Trifluoroacetate 10 inlet of the injection valve. A system backpressure of 9. Phosphate20 µS 15.9 MPa (2300 psi) is ideal. Confirm that the chro- 1 6 matographic pathway has no leaks. For more informa- 7 5 tion, see the Operator’s Manual for the EG50 Eluent 8 Generator System (Document Number 031908). 4 Using the Chromeleon workstation, turn on the EG50 to deliver the highest eluent concentration 0 0161412108642 81 20 required by the method. Allow the AS50 thermal Minutes compartment to stabilize at 30 °C. Determine the status 19446 of the system by measuring the short-term noise. Figure 1. Analysis of anion standard with TFA. Baseline noise should be less than 5 nS over a period of 5–10 min when measured in 1-min segments. It may take 12 h or more for the system to equilibrate to a The separation of TFA from the anions in the stable background conductivity for trace analysis. When samples was achieved with a series of eluent concentra- performing trace analysis, we recommend running the tion step changes. The method begins with an initial system overnight to equilibrate for use the following day. eluent concentration of 22 mM KOH to elute weakly retained ions such as fluoride, acetate, and formate. An RESULTS AND DISCUSSION eluent step change to 28 mM KOH at 6 min separates An anion-exchange column for monitoring residual trifluoroacetate from the other matrix anions such as TFA in high-ionic-strength pharmaceutical buffers sulfate and nitrate. The SRS current setting at the time should ideally have two characteristics. The column of separation of TFA is set at 80 mA. This current should have a sufficient ion-exchange capacity for the provides the optimum suppression with the least high ionic matrix and should separate TFA from the baseline noise. After TFA has eluted, the eluent is step matrix anions that are present at high concentrations. changed to 50 mM KOH at 12 min to clean the column The IonPac AS18 column has both of these characteris- of any highly retained matrix anions such as phosphate. tics. The 4-mm AS18 set column has an anion-exchange Afterward, the eluent is stepped back to 22 mM KOH to capacity of 285 µeq and is an excellent match for the reequilibrate the column for the next injection. target application. TFA is well resolved from the early- eluting anions under optimized conditions. Figure 1 shows the AS18 separation of an anion standard that includes TFA.

44 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography 50 mM Column: IonPac AG18 and AS18 Peaks: 1. Chloride– 50 Eluent:22 mM KOH for 0–6 min, 2. Trifluoroacetate300 ng/mL KOH Eluent Gradient Program 28 mM KOH for 6–12 min, 3. Phosphate – 50 mM KOH for 12–15 min, 22 mM KOH for 15–20 min Sample: Phosphate-buffered saline mM Eluent Source: EG50 1 mM potassium phosphate 155 mM sodium chloride Eluent Flow Rate: 1.0 mL/min 28mM 3 mM sodium phosphate Oven Temp.: 30 °C 22 mM 22 mM Suppressor: ASRS, AutoSuppression, 20 recycle mode ASRS ULTRA Current: 80 mA for 0–14 min, 160 124 mA for 14.01–17 min, SRS Current Program 80 mA for 17.01–20 min 124 mA Inj. Volume: 100 µL

mA 10 1 3 80 mA 80 mA

60 015 10 5 02 Minutes 19448A 2 Figure 2. IC method for determination of TFA using the IonPac µS AS18 and RFIC.

This optimized eluent step change method was quickly developed using the RFIC system. By simply programming the EG50 Eluent Generator, a number of 0 isocratic methods and a variety of step-gradient eluent 0161412108642 81 20 methods were quickly evaluated. This evaluation did not Minutes 19449 require the preparation of different eluents to achieve the ideal proportioning for proper method evaluation. A Figure 3. Determination of trace TFA in phosphate-buffered saline using the IonPac AS18 and RFIC. higher ASRS current setting of 125 mA is applied at 14 min to account for the higher eluent concentration used. A delay of approximately 2 min occurs for the using a 100-µL injection. This separation is challenging higher eluent concentration to reach the suppressor after because of the excess presence of matrix components that the eluent concentration change command is given to elute before and after the detection of a trace amount of the EG50. This delay is mainly due to the column void TFA. We recommend using the detection parameter in volume. In contrast, the change in current setting to the Chromeleon software called “Inhibit Integration” to ensure ASRS shows an immediate response. By applying the the accurate determination of TFA. Setting the “Inhibit optimal current to the suppressor, it is possible to Integration” command to “On” at 0.00 min stops the achieve low noise at every point of the separation. integration from the beginning of the run. It remains off Separations are performed at 30 °C to provide the best until the command to turn the “Inhibit Integration” retention time reproducibility. Figure 2 illustrates the command “Off” is set at 11.0 min. This setting starts eluent gradient program and ASRS current program for integration until “Inhibit Integration” is turned “On” again the method. at 13.5 min. Integration will be inhibited from this point This method was applied to the determination of until the end of the run. Because TFA is the only analyte of TFA in the following samples: a phosphate-buffered interest, this detection parameter greatly simplifies detec- saline (PBS),1 two buffers used in a recombinant protein tion and data processing. The integration window for recovery process,4 and a commercial peptide.2,3 Figure 3 stopping and starting detection should be modified shows the determination of 300 ng/mL of TFA in PBS according to the specific matrix of interest.

45 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography A calibration curve was obtained using TFA standards at 100, 300, and 1000 ng/mL prepared in Column:IonPac AG18 and AS18 Eluent: 22 mM KOH for 0–6 min, 28 mM KOH for 6–12 min, PBS. Three replicate injections were performed at each 50 mM KOH for 12–15 min, 22 mM KOH for 15–20 min concentration level. Results showed that TFA yielded a Eluent Source: EG50 2 Eluent Flow Rate: 1.0 mL/min linear response with a coefficient of determination (r ) Oven Temperature: 30 °C of 0.9979. The method detection limit (MDL) was Suppressor: ASRS, AutoSuppression, recycle mode estimated to be 100 ng/mL TFA in PBS by measuring a ASRS ULTRA Current: 80 mA for 0–14 min, 124 mA from 14.01–17 min, 80 mA for 17.01–20 min TFA peak three times higher than the background noise Inj. Volume: 50 µL (S/N = 3). An MDL was calculated using the standard Peaks: 1. Acetate/chloride deviation for seven replicate injections of 100 ng/mL 2. Trifluoroacetate 300 ng/mL TFA in the PBS.5 TFA was spiked at 100 ng/mL to be in the same concentration range as the estimated MDL and Sample:Protein Buffer #1 0.1 M acetic acid multiplied by the Student’s t value for the 99.5% 0.25 M sodium chloride confidence limit. The standard deviation was multiplied 0.01% Tween 20, pH3 1 mg/mL bovine serum albumin by the Student’s t value for the 99.5% confidence limit. A method detection limit for TFA was calculated to be 86 ng/mL in the PBS matrix under these conditions. 4 1 Recovery of TFA for a 300-ng/mL spike in PBS was 98.7% for (6) replicate injections (293 ± 2.7 ng/mL). The retention time of TFA was 11.7 ± 0.011 min with an RSD of 0.09%. These results compare favorably with the work by Fernando and coworkers using the IonPac µS AS11-HC column.1 They reported an MDL of 10 ng/mL

for TFA in PBS after reduction of the matrix chloride 2 concentration using an OnGuard® Ag pretreatment cartridge. We report an MDL of 86 ng/mL for TFA in PBS without a sample preparation step. This method is also applicable to monitoring TFA 1 in the buffers used in a recombinant protein recovery 015 10 5 02 process. Buffers were prepared according to the specifi- Minutes 19452 cation of the manufacturer (see the “Samples” section) with the addition of 1 mg/mL bovine serum albumin to Figure 4. Determination of trace TFA in protein buffer #1. simulate the presence of protein. The method was optimized using a 50-µL injection to give the best sensitivity for the determination of TFA in these two eluent.4 By using a higher injection volume (50 µL) with buffers. Recovery of TFA for a 300-ng/mL spike in this the higher-capacity IonPac AS18 column, we were able buffer was 98% for five replicate injections (296 ± 19 to achieve an MDL of 36 ng/mL for TFA. Figures 4 and ng/mL) based on a calibration curve prepared in the 5 show chromatograms for a 300 ng/mL TFA spike in matrix. The retention time of TFA was 12.7 ± 0.021 min both of these buffers. Linearity for TFA in these two with an RSD of 0.17%. Kabakoff and coworkers buffers yielded coefficients of determination (r2) greater reported an MDL of 300 ng/mL for TFA in this sample than 0.999. Recovery of TFA for a 50 ppb spike in with a 10-µL injection using a 4 × 250 mm IonPac AS14 Protein Buffer #2 was 115% for seven replicate injections. column (65 µeq/column capacity) with a carbonate-based

46 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography Column: IonPac AG18 and AS18 Column: IonPac® AG18 and AS18 Eluent: 22 mM KOH for 0–6 min, 28 mM KOH for 6–12 min, Eluent: 22 mM KOH for 0–6 min, 28 mM KOH for 6–12 min, 50 mM KOH for 12–15 min, 22 mM KOH for 15–20 min 50 mM KOH for 12–15 min, 22 mM KOH for 15–20 min Eluent Source: EG50 Eluent Source: EG50 Eluent Flow Rate: 1.0 mL/min Eluent Flow Rate:1.0 mL/min Oven Temperature: 30 °C Oven Temperature:30 °C Suppressor: ASRS, AutoSuppression, recycle mode Suppressor: ASRS, AutoSuppression, recycle mode ASRS ULTRA Current: 80 mA for 0–14 min, 124 mA for 14.01–17 min, ASRS ULTRA Current: 80 mA for 0–14 min, 124 mA for 14.01–17 min, 80 mA for 17.01–20 min 80 mA for 17.01–20 min Inj. Volume: 50 µL Inj. Volume: 50 µL

Peaks: 1. Chloride – Peaks:1.System peak –ng/mL 2. Trifluoroacetate 300 ng/mL 2. Acetate– 3. Formate– 2.0 2 Sample: Protein Buffer #2 4. Chloride 15 0.1 M Tris pH 7.4 5. Carbonate– 0.14 M sodium chloride 6. Unidentified – 0.01% Tween 20 7. Sulfate 140 1 mg/mL bovine serum albumin 8. Nitrate14 9. Trifluoroacetate100 4 10.Phosphate25 1 µS 5

1 3 10 7 9 8 4 6

µS

2 0.5 015 10 5 02 Minutes 19456

Figure 6. Determination of trace TFA in a commercial peptide: 1 40 µg/mL angiotensin II spiked with 100 ng/mL TFA. 015 10 5 02 Minutes 19454 peptide preparation. Figure 6 shows the peptide solution Figure 5. Determination of trace TFA in protein buffer #2. spiked with 100 ng/mL of TFA. Because this peptide was prepared as an acetate salt, a large acetate peak appears. Trace amounts of chloride and sulfate were This method was also applied to the determination also detected. A 50 ng/mL spike of TFA was completely of TFA in a commercial peptide. A solution of human recovered (101% for n = 7), demonstrating that the angiotensin II protein was prepared at 40 µg/mL in method is valid for determining TFA in this sample. deionized water with and without a spike of 100 ng/mL Table 1 summarizes the calibration results and calcu- of TFA. Both solutions were analyzed using the method lated method detection limits in the human angiotensin developed in this study. No TFA was detected in the II protein solution, including the three different buffers.

47 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography Table 1. Calibration Results and Calculated MDLs in the Human Angiostensin II Protein Solution Including the Three Different Buffers

Matrix Data points** r2 Dynamic Method DetectionStandard Used to Range Limit (MDL)* Calculate MDL (ng/mL) (ng/mL) (ng/mL)

Phosphate Buffered Saline 9 0.9979 100–1000 86 100 Protein Buffer #1 9 0.9997 100–1000 36 100 2# reffuB nietorP reffuB 2# 9 0001–0016899.0 51 03 editpeP laicremmoC editpeP 9 7999.0 003–03 4 01

* MDL = (S.D.) × (ts) 99.5%, where (ts) is for a 99.5% single-sided Student’s t test distribution for n = 7 ** Three concentrations injected in triplicate

REFERENCES SUPPLIERS 1. Fernando, P. N.; McLean, M. A.; Egwu, I. N.; Fisher Scientific, 2000 Park Lane, Pittsburgh, PA deGuzman, E.; Weyker, C. J. Chromatogr., A 2001, 15275-1126 USA, Tel: 800-766-7000, 920, 155–162. www.fishersci.com. 2. Dionex Corporation. Application Note 115; Fluka Chemika-BioChemika, Fluka Chemie AG, Sunnyvale, CA. Industriestrasse 25, CH-9471, Buchs, Switzerland, 3. Simonzadeh, N. J. Chromatogr. 1993, 634, Tel: +81 755 25 11, www.sigma-aldrich.com. 125–128. Upchurch Scientific, 619 West Oak Street, P.O. Box 4. Kabakoff, B.; Blank, A.; Heinsohn, H. Final 1529, Oak Harbor, WA 98277-1529 USA, Program, International Ion Chromatography Tel: 1-800-426-0191, www.upchurch.com. Symposium, Santa Clara, CA, Sept 14–17, VWR Scientific Products, 1310 Goshen Parkway, West 1997; 165. Chester, PA 19380 USA, Tel: 800-932-5000, 5. Glaser, J.; Foerst, G.; McKee, G.; Quave, S.; Budde. www.vwr.com. W. Environ. Sci. Technol. 1981, 15(12) 1426.

48 Determination of Residual Trifluoroacetate in Protein Purification Buffers and Peptide Preparations by Reagent-Free Ion Chromatography Application Note 164 Application Note 164

Assay for Citrate and Phosphate Assay for Citrate and Phosphate in Pharmaceutical in Pharmaceutical Formulations Formulations Using Ion Chromatography Using Ion Chromatography

INTRODUCTION Citrate has been successfully separated by ion- Citric acid is a common ingredient found in many exchange,4,5 ion-exclusion,6,7 and reversed-phase8 liquid pharmaceutical formulations. The most common use of chromatography in a wide range of sample matrices, citric acid in pharmaceuticals is the effervescent effect it including those of pharmaceutical and biological origin. produces when combined with carbonates or bicarbon- The most common reported detection of these separa- ates in antacids and dentifrices. Citrate is also widely tions is indirect UV absorbance; however, conductivity used as a flavoring and stabilizing agent in pharmaceuti- and refractive index detection have also been used. cal preparations to mask the taste of medicinal flavors. Separation of citric acid by reversed-phase liquid Citric acid can act as a buffering agent and assist in the chromatography requires a low mobile phase pH to dispersion of suspensions to help maintain stability of inhibit the ionization of citric acid.8 Furthermore, ion- the active ingredients1 and improve the effectiveness of exclusion separations generally have long retention antioxidants.2 It may also be used as an anticoagulant to times for citric acid unless an organic modifier is used.7 preserve blood for transfusion and as an ingredient of Because citrate is a very poor absorbing analyte, a rectal enemas.2 mobile phase with a strong UV-absorbing chromophore The United States Pharmacopeia (USP) has adopted is required for indirect UV detection.9 Chalgari and Tan several different assays for citrate in various pharmaceu- described a citrate assay for some pharmaceutical tical dosage forms. These analytical techniques include dosage forms with USP monographs that uses ion calorimetry, gravimetry, ion-exclusion chromatography, chromatography (IC) with indirect photometric detec- and reversed-phase liquid chromatography.3 Method tion.10 This method used trimesic acid, a UV-absorbing variation is usually required for many of these tech- eluent, as the mobile phase to detect citrate as a negative niques to assay a specific dosage form. This method peak at a wavelength of 280 nm. However, the method variation results in the use of different color-forming required proper pH adjustment of the mobile phase with reagents, mobile phases, columns, and detectors. For NaOH to produce consistent retention times. The instance, a dosage form containing citrate and phos- retention time of citric acid decreases as its ionization phate requires the use of pyridine and acetic anhydride decreases at low pH values (pH 3.2–4.5) and increase at for the determination of citrate, and ammonium molyb- higher pH values (pH 4.5–6.0) as ionization increases.11 date, hydroquinone, and sodium sulfite for a separate determination of phosphate. The prescribed assays are time consuming, labor intensive, require extensive analyst training, and may yield significant measurement errors.

49 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography IC with suppressed conductivity detection is the REAGENTS AND STANDARDS chromatographic technique of choice for citrate deter- Deionized water, Type I reagent-grade, 18 MΩ-cm minations.12 In addition, IC can simultaneously deter- resistivity or better mine phosphate and other anions that are present in Citric acid (USP, Catalog #1134368) some pharmaceutical formulations and uses eluents that Sodium dihydrogen phosphate monohydrate, do not require expensive reagents or pH adjustments. NaH2PO4·H2O (EM Science) Aliphatic carboxylic acids, such as citric acid, generally Calcium chloride dihydrate, CaCl ·2H2O (Fisher Scientific) exhibit high affinities for anion-exchange stationary Sodium acetate anhydrous (Fluka Chemical Co.) phases. Thus, low-ionic-strength carbonate/bicarbonate buffer solutions are typically not suitable as eluents. Sodium chloride (J. T. Baker) However, when hydroxide eluents are used, citric acid Magnesium chloride hexahydrate, MgCl2·6H2O can be easily eluted from the column.13 (Sigma-Aldrich) In this application note, we report on the validation Sodium citrate dihydrate (Sigma-Aldrich) of an IC method for the determination of phosphate and Potassium chloride (Sigma-Aldrich) total citric acid in pharmaceutical formulations with a Sodium hydroxide, 50% (J. T. Baker) hydroxide-selective, anion-exchange column and suppressed conductivity detection. The method incorpo- rates an electrolytic eluent generator to automatically CONDITIONS produce a simple isocratic potassium hydroxide eluent, Columns: IonPac AS11 Analytical, 4 × 250 mm allowing the separation of phosphate and citrate on an (Dionex P/N 044076) IonPac® AS11 column in less than 10 min. The results IonPac AG11 Guard, 4 × 50 mm indicate that this method can replace 18 USP mono- (Dionex P/N 044078) graphs for the assay of citric acid or phosphate in Eluent: 20 mM potassium hydroxide USP 27-NF 22. The method was evaluated in terms of Eluent Source: ICS-2000 EG with CR-ATC linearity, precision, accuracy, ruggedness, and limit of Flow Rate: 2.0 mL/min quantitation for phosphate and citrate. Temperature: 30 °C Injection: 10 µL EQUIPMENT Detection: Suppressed conductivity, A Dionex ICS-2000 Reagent-Free™ Ion Chromatography ® (RFIC) System was used in this work. The ICS-2000 ASRS ULTRA II, 4 mm is an integrated ion chromatograph and consists of: (Dionex P/N 061561) AutoSuppression® recycle mode Eluent generator 100 mA current Column heater System Pump with degas Backpressure: ~2300 psi EluGen® EGC II KOH Cartridge Run Time: 10 min (Dionex P/N 058900) CR-ATC (Dionex P/N 060477) AS50 Autosampler Chromeleon® 6.5 Chromatography Workstation This application note is also applicable to other RFIC systems.

50 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography PREPARATION OF SOLUTIONS AND STANDARDS SAMPLE PREPARATION All liquid samples should be appropriately diluted Eluent Solution with DI water so that the concentration of citrate and Generate 20 mM KOH eluent on-line by pumping phosphate fit within the calibration range. For solid deionized (DI) water through the ICS-2000 EG device. citrate samples, such as potassium citrate extended- Set the eluent concentration using Chromeleon release tablets that contain insoluble components, a software or from the front LCD panel of the ICS-2000. portion equivalent to ~100 mg citric acid (powdered Chromeleon Chromatography Management Software form) was added to 300 mL of hot DI water (~80 °C) tracks the amount of KOH used and calculates the and magnetically stirred for ~30 min while maintaining remaining lifetime of the EGC II KOH cartridge. the temperature between 70–80 °C. The solution was Alternatively, prepare 20 mM NaOH by pipetting allowed to cool and then transferred to a 500-mL 1.05 mL of 50% (w/w) aqueous NaOH from the reagent volumetric flask and diluted to volume with DI water to bottle into a 1.00-L volumetric flask containing about prepare the sample stock solution. For completely 500 mL of degassed DI water. Bring to volume with soluble solid samples containing citrate (e.g., efferves- degassed DI water, mix, and degas by sparging with cent tablets), a finely ground portion equivalent to helium or sonicating under vacuum for 10 min. Atmo- ~100 mg citric acid should be added to 300 mL of spheric carbon dioxide readily dissolves in dilute basic DI water in a 500-mL volumetric flask and diluted to solutions, forming carbonate. Carbonate contamination the mark to prepare the sample stock solution. In this of eluents can affect analyte retention times, resulting in study, most samples were diluted 1000-fold for citrate performance that is not equivalent to electrolytically determinations and approximately 200-fold for phos- producing the hydroxide eluent on-line using an eluent phate determinations. generator. Store the eluent in plastic labware. Maintain an inert helium atmosphere of 3–5 psi in the eluent SYSTEM PREPARATION AND SETUP reservoir to minimize carbonate contamination. Prepare the ASRS ULTRA II for use by hydrating the suppressor. Use a disposable plastic syringe and Stock Standard Solutions An official USP citric acid reference standard was push approximately 3 mL of degassed DI water through the “Eluent Out” port and 5 mL of degassed DI water dried in an oven at 105 °C for 2 h immediately before use. To prepare a 500-mg/L citric acid stock standard, through the “Regen In” port. Allow the suppressor to sit weigh exactly 250 mg of the dried citric acid, add to a for approximately 20 min to fully hydrate the suppres- 500-mL volumetric flask, and dilute to volume with sor screens and membranes. Install the ASRS ULTRA II DI water. To prepare a mixed citrate/phosphate stock for use in the recycle mode according to the Installation and Troubleshooting Instructions for the ASRS ULTRA II standard with 300 mg/L phosphate (as NaH2PO4·H2O), (Document No. 031956). weigh 150 mg NaH2PO4·H2O, add to a 500-mL volumetric flask containing 250 mg citric acid, and Install the EGC II KOH cartridge in the ICS-2000 dilute to the mark with DI water. Store in polypropylene and configure it with Chromeleon chromatography software. Condition the cartridge as directed by the bottles at 4 °C. EGC II Quickstart (Document No. 031909) for 30 min Working Standard Solutions with 50 mM KOH at 1 mL/min. Install a 4 × 50 mm Prepare working standard solutions at lower IonPac AG11 and 4 × 250 mm IonPac AS11 column. concentrations by adding an appropriate amount from Make sure that the system pressure displayed by the the stock standard solutions and 5 mL of 20 mM NaOH pump is at an optimal pressure of ~2300 psi when 20 mM to a 100-mL volumetric flask. Dilute to the mark with KOH is delivered at 2.0 mL/min so the degas assembly DI water. The 20 mM NaOH solution used for standard can effectively remove hydrolysis gas from the eluent. If and sample preparation should be prepared fresh daily. necessary, install additional backpressure tubing supplied with the ICS-2000 shipping kit to adjust the pressure to 2300 ± 200 psi. Because the system pressure can rise over time, trim the backpressure coil as necessary to maintain a system pressure between 2100–2500 psi.

51 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography RESULTS AND DISCUSSION In general, highly charged analytes, such as citrate, Column: IonPac AS11, 4 × 250 mm IonPac AG11, 4 × 50 mm are difficult to elute from most anion-exchange columns Eluent: 20 mM potassium hydroxide without using a concentrated eluent. To reduce the Eluent source: ICS-2000 EG with CR-ATC Flow rate: 2.0 mL/min elution time, the eluent anion should have a high Temperature: 30 °C selectivity for the resin. Therefore, an anion-exchange Inj. volume: 10 µL Detection: Suppressed conductivity, column with a high selectivity toward hydroxide eluent, ASRS ULTRA II, 4 mm, in combination with a low anion-exchange capacity of AutoSuppression recycle mode, 45 µeq/column, was chosen to assay for citric acid. This 130 mA column allows the separation of a wide range of inor- 3.5 Peaks: AB 1. Phosphate 12 2.6 mg/L ganic and organic anions, including polyvalent ions, 2. Citrate2021 such as citrate, using a low to moderate eluent concen- A 2 tration. In addition, a hydroxide eluent has the following 1 advantages: (1) it is readily available, (2) it can be µS electrolytically generated at an appropriate concentra- tion, and (3) it is suppressed to water to yield an exceptionally low background conductance and lower noise, therefore improving the limits of detection and 0.5 quantitation. An electrolytically generated potassium 3.5 hydroxide eluent was used for the separation of phos- phate and citrate in different pharmaceutical formula- 2 tions to increase method automation and allow easy B method transfer between laboratories. The assay was µS evaluated in terms of linearity, limit of quantitation, specificity, precision, accuracy, and ruggedness. 1 All calibration standards used in this assay were prepared in 1 mM NaOH. A total of 12 calibration data 0.5 sets were acquired using either combined citric acid and 10.08.87.56.35.03.82.51.30.0 Minutes phosphate standards or standards containing only citric 20499 acid. A calibration curve was generated with citrate in Figure 1. Separation of phosphate and citrate on the IonPac AS11 the range of 2–100 mg/L using seven concentration with (A) standard and (B) assay for citrate in an anticoagulant levels for the combined standard to assay formulations citrate, phosphate, dextrose, and adenine dosage form. containing citrate and phosphate, and in the range of 5–70 mg/L using five concentration levels for assay of The USP compendial method for validation <1225> dosage forms containing only citrate. A calibration specifies a signal-to-noise (S/N) ratio of 10 for the curve was generated with phosphate in the range of determination of the limit of quantitation (LOQ).14 The 1.2–60 mg/L. Each calibration was linear over the baseline noise was determined by measuring the peak- specified range using a least-squares regression curve to-peak noise in a representative 1-min segment of the with correlation coefficient (r2) of 0.9990 or better. baseline where no peaks elute. Typical baseline noise Figure 1A shows a chromatogram of a combined for this method using the ASRS ULTRA II suppressor phosphate and citrate standard separated on an IonPac in the recycle mode is ~2 nS/min. The determined LOQ AS11 and Figure 1B shows the same analytes analyzed for phosphate and citric acid over three consecutive in an anticoagulant citrate, phosphate, dextrose, and days was approximately 0.2 mg/L (S/N = 10). The adenine dosage form. method validation did not require a determination of the detection limit. Table 1 summarizes the calibration and LOQ data.

52 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography The USP defines ruggedness as a measure of the Table 1. Summary of Calibration and Limit of reproducibility of the method obtained by the analysis Quantitation Data for Citrate and Phosphate of the same samples under a variety of conditions. Coefficient of Ruggedness is typically expressed as the lack of influ- Concentration Determination ence on the assay results under different conditions that a a Range Range LOQ would normally be expected from laboratory to labora- Analyte (mg/L) (r2) (mg/L) tory and from analyst to analyst when operating under Citrate 2–100 0.9993–0.9994 0.20 the specified method parameters. We evaluated rugged- Citrate 5–70 0.9990–0.9998 0.20 ness of the method by using analysts, instruments, batch lots of the same column, and eluent preparation proce- Phosphate 1.2–60 0.9999 0.20 dures as variables. The precision was determined by a For three independent calibrations (3 days) using the average measured concentrations (based on duplicate injections) and calculating the RSDs for the separate assays. Table 3 shows the precisions for each Method performance was measured in terms of the variable tested for both an individual analyst and for precision of replicate sample injections and recovery of both analysts. For analyst A, the RSDs ranged from spiked samples. The relative standard deviations (RSDs) 0.17–2.00% compared to 0.31–2.60% for analyst B. The of the measured peak areas were calculated for the greatest total RSD for both analysts was ~2.40%. Based target analytes prepared at a concentration of ~20 mg/L on these data, the method was found to be rugged in citrate and 12 mg/L phosphate. The intraday precision terms of the variables investigated. (i.e., a sequence of consecutive injections) for citrate Table 4 summarizes the results from the assay of and phosphate from independently prepared solutions nine different pharmaceutical formulations for citric analyzed on three separate days was <1% for citrate and acid and phosphate (if present). The same samples were <0.5% for phosphate on each day. The interday (i.e., day analyzed on three consecutive days using independently to day) precision for a three-day period (i.e., three prepared standards and diluted dosage solutions. The independent sample preparations) for citrate was <2% calculated concentrations of these samples measured on and for phosphate <1%. Recoveries were determined by separate days was consistent with a maximum differ- adding known amounts of analyte to the sample solu- ence of ~2% from day to day. In most cases, the mea- tions. The calculated recoveries were within 95–105% sured values were very close to the label amounts and for all samples. Table 2 summarizes the precision and within the specifications according to their respective recovery data for citrate and phosphate. USP 27-NF 22 monographs. However, the assay for phosphate in the anticoagulant solution and the assay for citrate in the oral rehydration solution were outside the specifications established by the USP. The amounts Table 2. Accuracy and Precision for Citrate and stated on the labels are based on the average from 20 to Phosphate in Pharmaceutical Formulations 25 sample containers. In this applications note, the Intraday Interday Range of values are based on the assay of one or two bottles. Precision Range Precision Range Recoveries Also, the methods used to determine the label values are Analyte (% RSD) (% RSD) (%) based on current USP procedures that are significantly Citrate 0.16–0.91a 0.49–1.94a 95.3–105.1c different than the method presented in this application note. Therefore, the difference in the formulation label Phosphate 0.19–0.49b 0.41–0.51b 94.8–104.8d values and the experimental values may be due to a Precision range for eight samples analyzed on three separate days from independently method variation, variation in the precision of the prepared solutions containing citric acid current USP methods, and the differences in sample size b Precision range for two samples analyzed on three separate days from independently prepared solutions containing phosphate used to establish the label and experimental values.

c Added 1–2.5 mg/L citric acid to nine samples prepared at 100% of the target concentration

d Added 0.6–1.5 mg/L phosphate to two samples prepared at 100% of the target concentration

53 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography Table 3. Summary of Results from the Ruggedness Study

Analyst A Analyst B Total elpmaS etylanA (% RSD) (% RSD) (% RSD)

Anticoagulant citrate phosphate dextrose adenine solution for citrate assay Citrate 0.17 0.31 1.51 Anticoagulant citrate phosphate dextrose adenine solution for phosphate assay Phosphate 2.00 2.60 2.17 Citric acid, magnesium oxide, sodium carbonate irrigation solutionCitrate 0.93 1.60 2.39 stelbat esaeler dednetxe etartic muissatoP etartic dednetxe esaeler stelbat etartiC 56.1 46.0 27.1

Table 4. Comparison of the Citrate and Phosphate Concentrations Obtained by IC with Suppressed Conductivity Detection to the Label Amounts

Label Amount Experimental Averagec ± SD elpmaS etylanA (mg/mL)a (mg/mL)a

Anticoagulant citrate phosphate dextrose adenine solution for citrate assay Citrate 20.1721.18 ± 0.10 Anticoagulant citrate phosphate dextrose adenine solution for phosphate assayPhosphate 2.222.81 ± 0.010 Citric acid, magnesium oxide, sodium carbonate irrigation solutionCitrate29.6 29.9 ± 0.4 stelbat esaeler dednetxe etartic muissatoP etartic dednetxe esaeler stelbat etartiC 01 qem 3.01 ± 0.2 meq Anticoagulant citrate phosphate dextrose solution for citrate assay Citrate 20.17 20.79 ± 0.23 Anticoagulant citrate phosphate dextrose adenine solution for phosphate assayPhosphate 2.22 2.20 ± 0.02 noitulos laro etartic muisengaM etartic laro noitulos TLNetartiC b 75.9 86.9 ± 1.8 noitulos laro dica cirtic dna etartic muidoS etartic dna cirtic dica laro noitulos etartiC 4.621 3.821 ± 1.6 stelbat tnecsevreffe dica cirtic dna etanobracib muidoS etanobracib dna cirtic dica tnecsevreffe stelbat 0001etartiC 7.4401bat/gm ± 21.5 mg/tab 2 epyt noitcejni setylortcele elpitluM setylortcele noitcejni epyt 2 etartiC 315.0 715.0 ± 0.003 noitulos noitardyher larO noitardyher noitulos etartiC 29.1 55.2 ± 0.05

a Amounts expressed as mg/mL citric acid or NaH2PO4·H2O unless otherwise noted b NLT = Not less than c Average and standard deviation of three independent determinations, two injections per day

CONCLUSION assays for 18 pharmaceutical formulations for citrate or An IC method using a low-capacity, hydroxide- citrate/phosphate in USP monographs. Laboratories that selective, anion-exchange column with suppressed currently support multiple USP citrate assays for conductivity detection provides an efficient and rapid different pharmaceutical formulations may be able to separation of phosphate and citrate in different pharma- standardize on this single IC assay. This assay can ceutical formulations. This method meets USP perfor- incorporate electrolytic generation of the potassium mance requirements in terms of specificity, linearity, hydroxide eluent to enhance the consistency of the precision, and recovery of samples spiked with phos- results between analysts and laboratories. phate and citrate. There are currently nine different USP

54 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography REFERENCES 11. Walker, T. A. J. Pharm. Biomed. Anal. 1995, 13, 1. Apac Chemical Corporation. Citric Acid Product 171–176. Information, 2002; http://www.apacchemical.com/ 12. Singh, R. P.; Smesko, S. A.; Abbas, N. M. CitricAcid.htm. J. Chromatogr., A 1997, 774, 21–35. 2. Chalgeri, A.; Tan, H. S. I. J. Pharm. Biomed. Anal. 13.Weiss, J. Ion Chromatography. 2nd ed.; VCH 1993, 11, 353–359. Publishers, Inc.: New York, 1995; pp. 126–127. 3. Bhattacharyya, L. Presented at the International Ion 14. United States Pharmacopeia 27 National Formulary Chromatography Symposium, Baltimore, MD, 22, General Chapter <1225>, Validation of October, 2002. Compendial Methods, U.S. Pharmacopeial Conven- 4. Holden, A. J.; Littlejohn, D.; Fell, G. S. J. Pharm. tion, Inc., Rockville, MD, 2004, 2622–2625. Biomed. Anal. 1996, 14, 713–719. 5. Lu, S.; Sun, X.; Shi, C.; Zhang, Y. J. Chromatogr., A SUPPLIERS 2003, 1012, 161–168. U.S. Pharmacopeia, 3601 Twinbrook Parkway, 6. Karmarkar, S.; Koberda, M.; Momani, J.; Kotecki, Rockville, MD 20852 USA, Tel.: 800-227-8772, D.; Garber, R. Presented at the International Ion www.usp.org. Chromatography Symposium, San Diego, CA, VWR Scientific Products, 1310 Goshen Parkway, West October, 2003. Chester, PA 19380 USA, Tel: 800-932-5000, 7. Chen, Q.; Mou, S.; Liu, K.; Yang, Z.; Ni, Z. www.vwr.com. J. Chromatogr., A 1997, 771, 135–143. Sigma-Aldrich Chemical Co., P.O. Box 14508, St. 8. Khaskahili, M. H.; Bhanger, M. I.; Khand, F. D. Louis, MO 63178 USA, Tel: 800-325-3010, J. Chromatogr., B 1996, 675, 147–151. www.sigma-aldrich.com. 9. Jenke, D. R. J. Chromatogr. 1988, 437, 231–237. Fisher Scientific, 2000 Park Lane, Pittsburgh, PA 10. Chalgeri, A.; Tan, H. S. I. J. Pharm. Biomed. Anal. 15275-1126 USA, Tel: 800-766-7000, 1996, 14, 835–844. www.fishersci.com.

55 Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography Application Note 200

Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer

INTRODUCTION 9 Urea is commonly used in protein purification, detection. This method was used to evaluate including large scale purification of recombinant proteins the efficiency of cyanate scavengers and make for commercial purposes, and in recombinant protein recommendations for protein buffers that reduced manufacturing to denature and solubilize proteins.1–3 Some cyanate accumulation and subsequent carbamylation proteins are readily soluble and denature at moderate reactions. The authors evaluated separate 0.1 M citrate, urea concentrations (4–6 M), however, most solubilize phosphate, and borate buffers in 8 M urea across a and denature at higher concentrations (8–10 M).4–6 Urea pH range of 5–9. They reported that the citrate (pH = 6) degrades to cyanate and ammonium in aqueous solutions. buffered urea solution demonstrated the best suppression The maximum rate of cyanate production occurs near of cyanate accumulation (<0.2 mM cyanate). However, neutral pH, the typical pH range of biological buffers.7 phosphate buffers at pH 6 and 7 (<0.5 mM cyanate) Cyanate can carbamylate proteins through a reaction were preferred over the citrate buffers because citrate with free amino, carboxyl, sulfhydryl, imidazole, phenolic actively carboxylates proteins. (It is sometimes used as hydroxyl, and phosphate groups.8 These are unwanted a carboxylating agent.) While this analytical method modifications that can alter the protein’s stability, was effective, the authors believed it could be improved function, and efficiency. While some of these reactions by using a high-capacity hydroxide-selective anion- can be reversed by altering the pH of the solution, exchange column with better chloride-cyanate resolution. cyanate-induced carbamylation reactions to N-terminal Hydroxide eluent delivers better sensitivity than amino acids, however–such as arginine and lysine–are carbonate/bicarbonate. Improved resolution between irreversible.8 Urea solutions are commonly deionized chloride and cyanate, combined with the capability to remove cyanate for this reason. Unfortunately, high to inject more concentrated samples due to the higher cyanate concentrations can accumulate in urea solutions column capacity also provides increased method regardless of prior deionization, with some urea buffers sensitivity. reporting cyanate concentrations as high as 20 mM.7 This application note shows determination of cyanate An accurate, sensitive method for measuring cyanate in in samples of 8 M urea, 8 M urea with 1 M chloride, and high ionic strength matrices is needed to help monitor and 8 M urea with 1 M chloride and 50 mM phosphate buffer ™ control quality in these buffers. (pH = 8.4) using a Reagent-Free Ion Chromatography ™ Ion chromatography (IC) is an ideal method for (RFIC ) system. This method provides improved cyanate determination. A 2004 publication showed sensitivity, allows smaller volume sample injections, determination of cyanate in urea solutions by IC using a lowers the required dilution, and demonstrates higher ® resolution of cyanate from chloride compared to the Dionex IonPac AS14 column with 3.5 mM Na2CO3/ previously published method. Cyanate is separated using 1.0 mM NaHCO3 eluent, and suppressed conductivity

56 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System a hydroxide-selective IonPac AS15 (5-µm) column with Sodium chloride (NaCl) (VWR, JT3624) electrolytically generated 25 mM potassium hydroxide Sodium cyanate (NaOCN) (Aldrich, 185086) eluent and suppressed conductivity detection. The Sodium phosphate, dibasic anhydrous (Na2HPO4) IonPac AS15 column is a 3 × 150 mm high-capacity (VWR, JT3828) (60 µEq/column) column, with a smaller particle size, Sodium phosphate, monobasic monohydrate diameter, and length than the IonPac AS14 column (NaH2PO4∙H2O) (VWR, JT3818) described in a previous method. These changes improve Urea (H2NCONH2) (VWR, JT4204-1) sensitivity, reduce eluent consumption, and allow higher sample throughput. The mobile phase is electrolytically Urea Matrix Sample Solutions generated, which reduces labor, improves consistency, and Urea solutions were prepared from solid compounds provides the reproducibility of an RFIC system. Linearity, in 18.2 MΩ-cm deionized water and diluted 50× prior to limit of detection (LOD), precision, recoveries, and cyanate determinations: stability of cyanate in urea as a function of temperature 8 M urea are discussed. 8 M urea with 1 M chloride 8 M urea with 1 M chloride and 50 mM phosphate EXPERIMENTAL (pH = 8.4) Equipment All solutions containing urea were prepared the same Dionex ICS-3000 RFIC-EG™ system consisting of: day of the experiments unless otherwise stated. SP Single gradient pump DC Detector and Chromatography module, single or CONDITIONS dual temperature zone configuration Columns: IonPac AS15 5-µm Analytical CD Conductivity Detector (3 × 150 mm, P/N 057594) EG Eluent Generator IonPac AG15 5-µm Guard, AS Autosampler with Sample Tray Temperature (3 × 30 mm, P/N 057597) Controlling option and 1.5 mL sample tray Eluenta: 25 mM KOH EluGen® EGC II KOH cartridge (P/N 058900) Eluent Source: EGC II KOH with CR-ATC Continuously Regenerated Anion Trap Column Flow Rate: 0.5 mL/min (CR-ATC, P/N 060477) Column Temperature: 30 ºC Chromeleon® 6.8 Chromatography Workstation Tray Temperature: 4 ºCb Virtual Column™ Separation Simulator (optional) Injection Volume: 5 µL Sample Vial kit, 0.3-mL polypropylene with caps and Detection: Suppressed conductivity, septa (P/N 055428) ASRS® 300 (2 mm, P/N a This application can be performed on other Dionex 064555), recycle mode, 31 mA RFIC-EG systems. Background Conductivity: <1 µS REAGENTS AND STANDARDS Baseline Noise: <2 nS Reagents System Backpressure: ~2200 psi Deionized water, Type 1 reagent-grade, 18.2 MΩ-cm Run Time: 22 min resistivity, freshly degassed by ultrasonic agitation and a Add a step change at 12 min to 65 mM KOH when applied vacuum. determining cyanate in samples containing phosphate. Use only ACS reagent grade chemicals for all The eluent conditions are: 25 mM KOH for 0 to 12 min, reagents and standards. and 65 mM KOH from 12 to 20 min, recycle mode, 81 mA. Chloride standard (1000 mg/L; Dionex, P/N 037159) b Temperature was maintained at 4 °C using the Sodium carbonate, anhydrous (Na2CO3) (VWR, JT3602) temperature-controlled autosampler tray to minimize changes in cyanate concentration.

57 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System PREPARATION OF SOLUTIONS AND REAGENTS Prepare the combined 8 M urea, 1 M sodium chloride It is essential to use high quality, Type 1 water, with (NaCl, FW 58.44 g/mole) matrix sample solution in a resistivity of 18.2 MΩ-cm or greater, and it must be a similar manner using 48.05 g urea, 5.84 g sodium relatively free of dissolved carbon dioxide. Degas the chloride, and deionized water in a 100 mL volumetric deionized water using ultrasonic agitation with applied flask. vacuum. Prepare the combined 8 M urea, 1 M sodium chloride, and 50 mM phosphate (49.5 mM sodium

1 M Stock Cyanate Standard Solution phosphate monobasic (NaH2PO4∙H2O), 0.5 mM sodium

To prepare a 1 M stock cyanate standard solution, phosphate dibasic (Na2HPO4) matrix solution using dissolve 6.50 g of sodium cyanate (NaOCN, FW 65.01 g/mol) 48.05 g urea, 5.84 g sodium chloride, 0.683 g sodium with deionized water in a 100 mL volumetric flask and bring to phosphate monobasic (NaH2PO4∙H2O, FW 58.44 g/mole), volume. Gently shake the flask to thoroughly mix the solution. 0.007 g sodium phosphate dibasic (Na2HPO4, FW 141.96 g/mole), and deionized water in a 100 mL Primary and Working Cyanate Standard Solutions volumetric flask. Bring to volume, and mix thoroughly To prepare a 1.00 mM cyanate primary standard, (pH = 8.4). pipette 100 µL of the 1 M cyanate stock standard solution To prepare 100-, 50-, and 10-fold dilutions of the into a 100-mL volumetric flask, bring to volume with matrix sample solutions for the dilution experiments, deionized water, and shake the flask gently to mix. pipette 200, 400, and 2000 µL, respectively, of the matrix To prepare 1, 2, 4, 10, 20, 50, and 100 µM cyanate sample solution into a (tared) 20 mL polypropylene individual working standards, pipette 50, 100, 200, 500, scintillation vial and add deionized water until a total 1000, 2500, and 5000 µL, respectively of the 1.00 mM cyanate weight of 20.00 g is reached. primary standard solution into separate 50 mL volumetric flasks. Bring to volume with deionized water and shake PRECAUTIONS gently to mix. The urea solutions must be stored frozen at –20 or Prepare the 0.13, 0.25, and 0.50 µM cyanate detection -40 °C and defrosted before use or prepared fresh daily limit standards by diluting the 1.0 µM cyanate working as the cyanate concentrations increase over time and with standard using serial dilutions. For example, a portion of increased temperature. the 1.0 µM cyanate standard is diluted 50% to 0.50 µM cyanate. The 0.13 and 0.25 µM cyanate standards are SYSTEM SETUP prepared in a similar way from the 0.25 and 0.50 µM Refer to the instructions in the ICS-3000 Installation10 cyanate standards, respectively. and Operator’s11 manuals, AS Autosampler Operator’s Store all standards at 4 °C. Prepare the 1–20 µM manual,12 and column,13 and suppressor14 product manuals standard solutions weekly and the primary and stock for system setup and configuration. standard solutions monthly.

Urea Matrix Sample Solutions

To prepare 8 M urea (H2NCONH2, FW 60.06 g/mole) matrix sample solution, dissolve 48.05 g urea with deionized water in a 100 mL volumetric flask, dilute to volume, and mix thoroughly.

58 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System RESULTS AND DISCUSSION Column: I onPac® AG15 5-µm, Eluent Source: EGC II KOH Method Development and Optimization AS15 5-µm (3 mm) Temperature: 30 °C Eluent: 25 mM KOH Flow Rate: 0.50 mL/min The challenge in this application is to accurately Inj. Vol: 5 µL determine low concentrations of cyanate in a matrix with 0.58 Detection: ASRS® 300, recycle, 31 mA Sample Prep.: 50-fold dilution high concentrations of salts (up to 20,000× higher) and to 5 Peaks: 1-2. Unknown — µM resolve a cyanate peak eluting near a significantly larger 3. Chloride — chloride peak. In a previous study, cyanate was separated 4. Cyanate 2.3 5. Carbonate — on a Dionex IonPac AS14 column using carbonate/ µS bicarbonate eluent at a flow rate of 1.2 mL/min, with 9 3 suppressed conductivity detection. The authors reported 4 a 1.3 min retention time difference between chloride and 12 cyanate. However, cyanate is not fully resolved from 0.41 chloride in urea solutions containing 1 M chloride. The 0 510 15 2220 Minutes authors also reported that the chromatogram’s baseline 25298 was affected when 8 M urea solutions were injected. Figure 1. Determination of 2 µM cyanate by ion chromatography. As urea eluted through the column, the baseline of the chromatogram increased >2 µS at approximately 3 min, To refine the results obtained by Virtual Column then slowly drifted downward to the original baseline. simulator, the retention times of a 5 µL injection of 1 mM The cyanate detection limit corresponding to the undiluted cyanate, chloride, and carbonate were determined on an urea solution is high (200 µM). An RFIC system with a IonPac AS15 5-µm, 3 x 150 mm column, using hydroxide-selective column and electrolytically generated 20, 25, and 30 mM potassium hydroxide at 0.5 mL/min hydroxide eluent can be easily optimized to resolve and 30 °C. The experiments showed cyanate well resolved low concentrations of cyanate from high concentrations from chloride and carbonate by 3.2 min of chloride in urea samples. Electrolytically generated (Rs > 3) using either 20 or 25 mM potassium hydroxide, hydroxide eluent improves sensitivity and reduces and, as expected, retention times decreased with baseline noise. increasing eluent strength. 25 mM potassium hydroxide In order to optimize conditions and to minimize eluent was selected for this assay. Figure 1 demonstrates the time required to select the hydroxide-selective, good peak response and peak asymmetry for 2 µM high-capacity column with the highest resolution for cyanate separated on the IonPac AS15 5-µm column using cyanate, the authors used the Chromeleon Virtual Column electrolytically generated 25 mM potassium hydroxide at simulator to model separation conditions. Nitrite was 0.5 mL/min. used to model cyanate because it is not currently part During the initial method evaluation, the effects of the Virtual Column database and was chosen after of column overload using 8 M urea samples with and reviewing chromatograms in column manuals for various without spiked concentrations of 1 M sodium chloride high-capacity hydroxide-selective columns. (Only were tested. The chromatograms of undiluted 8 M urea these types of columns were considered because some showed a similar baseline disturbance (3 µS) from urea as urea-containing solutions contain molar concentrations reported in the literature9 (not shown). As urea elutes from of chloride and other anions that may overload low- the column, the baseline shifts upward at approximately capacity columns). Virtual Column simulator was used to 3 min then slowly back to the original baseline at 15 min. evaluate the separation of chloride, nitrite, carbonate, and All tested urea-containing solutions demonstrated similar phosphate with isocratic hydroxide eluents at results. The magnitude of the baseline shifts reduced with 30 °C. The simulator results demonstrated that the IonPac dilution (Figures 2–4). Urea contains minor unidentified AS15 column using 20 mM KOH provided the optimum ions that do not interfere with the cyanate peak and elute chloride/nitrite and nitrite/carbonate resolution with from the column in less than 20 min. 8 M urea samples

Rs > 3. The Virtual Column simulator proved to be a did not overload the column. To determine whether the valuable tool in helping accelerate method development molar concentrations of chloride expected in the samples by eliminating the time required to select a column and would cause column overloading, 20-, 50-, and 100-fold eluent suitable for this application. dilutions of 8 M urea solutions containing 1 M chloride

59 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System were injected. The chromatography of the 20-fold dilution column has similar capacity per column but 30% less samples (50 mM chloride) showed distorted chloride volume than the AS14 column. These differences provide peaks typical of column overloading. At least 50× improvement in peak response and sensitivity. This allows dilutions of 8 M urea with 1 M chloride are required to for a smaller sample (5 µL) to be injected, resulting in avoid this phenomenon and to obtain adequate resolution less column overload and longer column life. The limit of of chloride and cyanate. quantification, using 10× S/N, is 0.8 µM (1.0 µM, S/N 11.6). Method Qualification To qualify the cyanate method, linearity, system Determination of Cyanate in Urea and Urea-Containing noise, limit of quantification, and LOD were evaluated. Solutions The peak area response of cyanate was determined from The method was applied to 50-fold dilutions of 1 to 100 µM, using triplicate injections of the calibration 8 M urea, 8 M urea with 1 M chloride, and 8 M urea with standards. The linearity of cyanate peak area responses 1 M chloride and 50 mM phosphate buffer (pH 8.4). The was determined with Chromeleon software using a least authors determined the urea and urea-containing solutions squares regression fit. The resulting correlation coefficient had similar cyanate concentrations of 1.1 ± 0.1 µM in (r2) was 0.9993. 50× diluted solutions (Table 1). The cyanate peak was The peak-to-peak baseline noise was measured in well resolved from chloride and carbonate, and had good 1 min segments from 20 to 60 min without injecting a peak shape (Figure 2A). An acceptably small baseline sample. The noise was acceptably low (0.95 ± 0.13 nS drift (<0.15 µS) from urea was observed starting at (n = 3)). To determine the limit of detection, seven 2.4 min and ending at about 15 min. In urea-containing replicates of 0.13, 0.25, and 0.50 µM cyanate were solutions with molar concentrations of chloride, cyanate injected. The peak responses of cyanate were compared elutes on the base of the large chloride peak and is against the baseline noise using 3× S/N. The LOD was not fully resolved from chloride (Figure 3A). The 0.25 µM cyanate (S/N 3.01). These detection limits are chromatography is similar for urea-containing solutions significantly lower than previously reported (2 µM) using with chloride and phosphate (Figure 4A). an IonPac AS14 column with bicarbonate/carbonate To determine the method accuracy, multiple additions eluents.9 This improvement is likely due to the advantages of cyanate (1.2, 2.2, and 3.6 µM) were added to 50× of using electrolytically generated hydroxide eluent with dilutions of 8 M urea samples. In addition, 50× dilutions suppressed conductivity detection. The IonPac AS15 of 8 M urea with 1M chloride and 8 M urea with

Table 1. Recoveries of Cyanate in Urea Solutions (Dilution Factor: 50×) Matrixa Amount RSD Amount Amount RSD Recovery Amount Amount RSD Recovery Amount Amount RSD Recovery Present Added Measured (%) Added Measured (%) Added Measured (%) (µM) (µM) (µM) (µM) (µM) (µM) (µM) A 1.13 0.50 1.20 2.26 0.64 96.7 2.20 3.32 1.38 99.7 3.62 4.92 0.87 103.6 B 1.11 1.07 0.91 1.88 1.03 93.1 1.31 2.43 1.44 100.4 2.22 3.48 0.55 104.4 C 1.00 1.41 0.81 1.60 1.10 89.5 1.24 2.27 0.69 101.1 1.70 2.66 1.16 98.0 A) 8 M Urea B) 8 M Urea, 1 M Chloride C) 8 M Urea, 1 M Chloride, 50 mM Phosphate Buffer pH = 8.4

60 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System 1 M chloride, and 50 mM phosphate buffer (pH = 8.4) Column: IonPac® AG15 5-µm, Temperature: 30 °C were similarly spiked with 0.9–2.2 µM, and 0.8–1.7 µM AS15 5-µm (3 mm) Flow Rate: 0.50 mL/min Eluent: 25 mM KOH Inj. Vol: 5 µL cyanate, respectively. The calculated recoveries were Eluent Source: EGC II KOH Detection: ASRS® 300, recycle, 31 mA >89% for all solutions (Table 1). The chromatograms of Sample Prep.: 50-fold dilution 0.85 Samples: A. 8 M urea the 50× dilution of 8 M urea and the same solution spiked B. Sample A with 2.2 µM of added cyanate with 3.6 µM cyanate are shown in Figure 2. Peaks 6–8 are unidentified ions from the urea matrix that elute from A B Peaks: 1-2. Unknown — — µM the column within 20 min, while run time was extended µS 3. Chloride — — to 22 min to ensure that these ions were fully eluted 5 4. Cyanate 1.1 3.1 3 5. Carbonate — — before the next injection. The chromatograms of unspiked 6-8. Unknown — — 4 2 and spiked with 2.2 µM cyanate in 50-fold dilution 1 B of 8 M urea 1 M chloride are shown in Figure 3. The A chromatograms of 8 M urea with 1 M chloride and 0.35 0 510 15 2220 50 mM phosphate buffer (pH=8.4), both unspiked and Minutes 25299 spiked with 1.7 µM cyanate and diluted 50× are shown in Figure 2. Comparison of A) 8 M urea B) 8 M urea with Figure 4. 2.2 µM cyanate. To determine the retention time and peak area precisions seven replicate injections of 2 µM cyanate were spiked into deionized water, 50× dilutions of Column: IonPac® AG15 5-µm, Flow Rate: 0.50 mL/min 8M urea, 8 M urea with 1M chloride, and 8 M urea with AS15 5-µm (3 mm) Inj. Vol: 5 µL Eluent: 25 mM KOH Detection: ASRS® 300, recycle, 31 mA 1M chloride and 50 mM phosphate buffer (pH 8.4). Eluent Source: EGC II KOH Sample Prep.: 50-fold dilution Cyanate had similar retention times for all samples— Temperature: 30 °C Samples: A. 8 M urea, 1 M chloride B. Sample A with 2.2 µM of 8.99 to 9.07 min (Table 2). The retention time and peak added cyanate 0.58 area precisions were <0.1 and <2 % for all three samples. 3 A B Peaks: 1-2. Unknown — — µM 3. Chloride 20,000 20,000 4. Cyanate 1.1 3.5 µS 5. Carbonate — — 5 6-8. Unknown — — Table 2. Retention Time and Peak Area 4 Precisions of 2 µM Cyanate Spiked Into 2 1 6 50-fold Dilution of Urea Solution B 7 8 a A Matrix Retention Time RSD Peak Area 0.35 (min) RSD 0 510 15 2220 Minutes 25300 Deionized Water 9.07 0.02 1.06 Figure 3. Comparison of A) 8 M urea 1 M chloride B) 8 M urea A 9.07 0.02 1.65 1 M chloride with 2.2 µM cyanate. B 9.03 0.04 0.65 C 8.99 0.06 1.69 Sample Stability A) 8 M Urea The accumulation of cyanate in urea as a function of B) 8 M Urea, 1 M Chloride 1,4–9 C) 8 M Urea, 1 M Chloride, 50 mM Phosphate Buffer pH = 8.4 temperature is frequently discussed in the literature. n = 7 To determine the stability of cyanate in urea over 4 days, a Freshly prepared solutions cyanate concentrations were determined from 50-fold dilutions of 8 M urea, 8 M urea with 1 M chloride, and 8 M urea with 1 M chloride and 50 mM phosphate buffer solutions (pH = 8.4). The solutions were stored at -40 °C, 4 °C (AS autosampler tray), and 25 °C during the four-day experiment. To elute phosphate in the phosphate buffered urea solution, the method was modified with a step change to 65 mM KOH after the carbonate peak at 12 min.

61 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System CONCLUSION Column: IonPac® AG15 5-µm, Eluent Source: EGC II KOH AS15 5-µm (3 mm) Temperature: 30 °C Urea degrades to cyanate, an unwanted contaminant Eluent: 25 mM KOH for 12 min, Flow Rate: 0.50 mL/min step change to 65 mM KOH, Inj. Vol: 5 µL in urea-containing buffers used for protein purification. 65 mM KOH from 12 to 20 min Detection: ASRS® 300, recycle, 31 mA Using a high-capacity anion-exchange column with Sample Prep.: 50-fold dilution Samples: A. 8 M urea, 1 M chloride suppressed conductivity detection, the authors accurately B. Sample A with 2.2 µM of determined low (µM) concentrations of cyanate in 3 7 added cyanate 0.85 50-fold dilutions of 8 M urea and urea solutions A B Peaks: 1-2. Unknown — — µM containing molar concentrations of chloride and mM 3. Chloride 20,000 20,000 4. Cyanate 1.0 2.3 concentrations of phosphate. This method allows fast, 5. Carbonate — — µS accurate determination of cyanate in urea-containing 6. Unknown — — 5 7. Phosphate 1,000 1,000 solutions. A Reagent-Free IC system ensures the highest 2 4 6 precision, eliminates the need to prepare eluents, and 1 B eliminates possible eluent preparation errors. A 0.35 0 510 15 20 Minutes 25301 REFERENCES Figure 4. Comparison of A) 8 M urea 1 M chloride and 50 mM 1. Holtham, S.B.; Schütz, F. The effect of cyanate on the phosphate buffer (pH = 8.4) to B) Sample A spiked with 1.2 µM stability of proteins. Biochim. Biophys. Acta, 1949, 3, cyanate. 65–81. 2. Hoylaerts, M.; Chuchana, P.; Verdonck, P.; Roelants, P.; Weyens, A.; Loriau, R.; De Wilde, M.; Bollen, A. Large scale purification and molecular 8 M Urea, Defrosted from –40 °C characterization of human recombinant 1-proteinase 8 M Urea, at 25 °C 8 M Urea, at 4 °C inhibitor produced in yeasts. J. Biotechol., 1987, 5, 1000 181–197. 900 800 3. Amersham Pharmacia Biotech. The Recombinant 700 600 Protein Handbook. Protein amplification and simple 500

Cyanate (µM) 400 purification. Edition AA, 18-1142-75. 4-7 Amersham 300 200 100 Pharmacia Biotech, Piscataway, NJ. 2000. 4–7. 0 0102030405060708090 100 4. Dirnhuber, P.; Schütz, F. The isomeric transformation h 25302 of urea into ammonium cyanate in aqueous solutions. J. Biochem., 1948, 42, 628–632. Figure 5. Effect of temperature on cyanate from urea solutions. 5. Marier, J.R.; Rose, D. Determination of cyanate, and a study of its accumulation in aqueous solutions of urea. Anal. Biochem., 1964, 7, 304–314. The experiments showed the total cyanate 6. Reh, G.; Spelzini, D.; Tubio, G.; Pico, G.; Farruggia, concentrations were stable in 8 M urea when stored at B. Partition features and renaturation enhancement –40 °C. However, cyanate concentration increased more of chymosin in aqueous two-phase systems. J. than 10-fold over four days when stored at 4 °C (from Chromatogr. B., 2007, 860, 98–105. 6 to 75 µM) and increased significantly when stored at 7. Hagel, P.; Gerding, J.J.T.; Fieggen, W.; Bloemendal, 25 °C (from 24 to 886 µM) (Figure 5). Cyanate had H. Cyanate formation in solutions of urea. I. similar stability in 8 M urea with 1 M chloride and Calculation of cyanate concentrations at different 8 M urea with 1 M chloride and 50 mM phosphate buffer temperature and pH. Biochim. Biophys. Acta, 1971, as in the 8 M urea solutions. Cyanate accumulation in 243, 366–373. urea was not inhibited by the phosphate buffer. In the 8. Bennion, B.J.; Daggett, V. The molecular basis previous study, the authors reported that scavengers–in for the chemical denaturation of proteins by urea. addition to phosphate and other buffers–were needed to Proceedings of National Academy of Science (PNAS), effectively suppress the accumulation of cyanate in urea.9 2003, 100, 5142–5147. These results agree with the previous study.

62 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System 9. Lin, M-F.; Williams, C.; Murray, M.V.; Conn, G.; SUPPLIERS Ropp, P.A. Ion chromatographic quantification of Fisher Scientific International Inc., Liberty Lane, cyanate in urea solutions: estimation of the efficiency Hampton, NH 03842 Tel: 1-800-766-7000 of cyanate scavengers for use in recombinant protein www.fisherscientific.com manufacturing. J. Chromatogr. B., 2004, 803(2), VWR International, Inc., Goshen Corporate Park West, 353–362. 1310 Goshen Parkway, West Chester, PA 19380 10. Dionex Corporation. Installation Instructions for Tel: 1-800-932-5000 www.vwrsp.com ICS-3000 Ion Chromatography System; Document Sigma-Aldrich, Inc., P.O. Box 951524, Dallas, TX Number 065032. Dionex Corporation, Sunnyvale, 75395-1524 Tel: 1-800-325-3010 CA. 2005. www.sigmaaldrich.com 11. Dionex Corporation. Operator’s Manual for ICS-3000 Ion Chromatography System; Document Number 065031. Dionex Corporation, Sunnyvale, CA. 2005. 12. Dionex Corporation. Operator’s Manual for AS Autosampler; Document Number 065051. Dionex Corporation, Sunnyvale, CA. 2005. 13. Product Manual for IonPac AG15 Guard and AS15 Analytical Columns; Document Number 031362. Dionex Corporation, Sunnyvale, CA. 2002. 14. Product Manual for the Anion Self-Regenerating Suppressor 300 and Cation Self-Regenerating Suppressor 300; Document Number 031956. Dionex Corporation, Sunnyvale, CA. 2007.

63 Direct Determination of Cyanate in a Urea Solution and a Urea-Containing Protein Buffer Using a Reagent-Free Ion Chromatography System Application Note 220

Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection

INTRODUCTION Ion chromatography (IC) with suppressed The U.S. Food and Drug Administration (FDA) is conductivity detection is a well-established technique responsible for protecting consumers by ensuring that for the determination of inorganic and organic ions in pharmaceuticals are safe by requiring the manufacturers pharmaceuticals.5–7 For the determination of anions, to verify their identity, strength, quality, and purity a hydroxide eluent is commonly used. Hydroxide is characteristics. Impurities that are present even in small suppressed to water, which provides exceptionally amounts may influence the safety and efficacy of the low background conductivity and baseline noise and, pharmaceutical product. According to the International therefore, very low detection limits. In Application Note Conference on Harmonization, impurities are defined as 190 (AN190), we demonstrated the determination of any component of the active pharmaceutical ingredient sulfate counter ion and anionic impurities in several water- (API) that is not the chemical entity defined as the API.1 soluble aminoglycoside antibiotics.8 Most of the samples Pharmaceutical impurities are categorized as organic, described in AN190 could be analyzed by direct injection inorganic, or residual solvents. Inorganic impurities after dilution with deionized water. In this Application that may be derived from the manufacturing process of Note (AN), we demonstrate the development of an IC bulk drugs include reagents, catalysts, ligands, heavy method for the determination of anionic impurities metals, and other materials (e.g., filter aids, charcoal).2 in a proprietary water-insoluble pharmaceutical. A For example, inorganic impurities may be present in the 2-mm IonPac® AS15 column with an electrolytically raw materials or may be derived from reagents, such generated potassium hydroxide eluent was used for the as phosphate buffers, used during the production of the determination of sub-mg/L concentrations of inorganic pharmaceutical. While the presence of many inorganic anion impurities in a proprietary pharmaceutical dissolved impurities at low concentrations have few toxicological in 100% MeOH. A 100-µg/L sample was concentrated consequences, significant variation in the impurity levels on an IonPac UTAC-ULP1 concentrator followed by from batch-to-batch can indicate that the manufacturing elimination of the MeOH matrix and pharmaceutical with process of the drug product is not adequately controlled.3,4 1 mL of deionized water to permit the determination of In most cases, these impurities should be removed or the target inorganic anions without matrix interferences. at least minimized in the final product. Therefore, the The linearity, detection limits, precision, and accuracy of identification, quantification, and control of impurities are the method are described. important during drug development in the pharmaceutical industry.

64 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection EQUIPMENT Eluent Source: EGC II KOH with CR-ATC Dionex ICS-3000 Reagent-Free™ Ion Chromatography Flow Rate: 0.40 mL/min (RFIC) system consisting of: Temperature: 30 °C (lower compartment) DP Dual Pump module (an SP Single Pump module 30 °C (upper compartment) can also be used) Inj. Voume: 100 µL EG Eluent Generator module Matrix Elim. Vol.: 1000 µL (DI water) DC Detector/Chromatography module (single or dual Concentrator: IonPac UTAC-ULP1, 5 × 23 mm temperature zone configuration) (P/N 063475) AS Autosampler with a 1-mL syringe (P/N 055066) CRD: CRD 200, 2-mm (P/N 062986) EluGen EGC II KOH cartridge (P/N 058900) Detection: Suppressed conductivity, ASRS® 300 Continuously-Regenerated Anion Trap Column, CR-ATC (2 mm), Recycle mode, 60 mA current (P/N 060477) System Chromeleon® 6.8 Chromatography Data System Backpressure: ~2400 psi Background REAGENTS AND STANDARDS Conductance: ~0.5-0.7 µS Deionized water, Type I reagent grade, 18 MΩ-cm Noise: ~1-2 nS/min peak-to-peak resistivity or better Run Time: 30 min Combined Seven Anion Standard, 100 mL *The column equilibrates at 10 mM KOH for 5 min (Dionex P/N 056933) prior to the next injection Fluoride Standard 1000 mg/L, 100 mL (Dionex P/N 037158 or Ultra Scientific, PREPARATION OF SOLUTIONS AND REAGENTS VWR P/N ULICC-003) Mixed Inorganic Anion Stock Solution Chloride Standard 1000 mg/L, 100 mL To estimate the concentration of the target anions in (Dionex P/N 037159 or Ultra Scientific, the sample, prepare a 1000-fold dilution of the Combined VWR P/N ULICC-002) Seven Anion Standard. Inject 100 µL of this standard followed by 1000 µL of deionized water. The separation Sulfate Standard 1000 mg/L, 100 mL should be similar to that shown in Figure 1. For this (Dionex P/N 037160 or Ultra Scientifc, application, nitrite and bromide were excluded from VWR P/N ULICC-006) the calibration standards because these anions were not Nitrate Standard 1000 mg/L, 100 mL (Ultra Scientific, detected in the sample or matrix blank. VWR P/N ULICC-004) Phosphate Standard 1000 mg/L, 100 mL (Ultra Scientifc, Stock Standard Solutions for Target Anions (1000 mg/L) VWR P/N ULICC-005) For several of the analytes of interest, 1000 mg/L Methanol, ACS grade (99.8% min), BDH standard solutions are available from Dionex or other (VWR P/N BDH1135-4LG) commercial sources. When commercial standards are not available, 1000 mg/L standards can be prepared CONDITIONS by dissolving the appropriate amounts of the required Columns: IonPac AG15 Guard, 2 × 50 mm analytes from ACS reagent grade salts (or better) in (P/N 053943) 100 mL of deionized water. Standards are stable for at least one month when stored at 4 °C. IonPac AS15 Analytical, 2 × 250 mm (P/N 053941) Eluent: 10 mM potassium hydroxide 0–8 min, 10 – 40 mM from 8 – 14 min, 40 – 60 mM from 14 – 20 min, 60 mM from 20 – 30 min*

65 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection SAMPLE PREPARATION Column: IonPac AG15, AS15, 2 mm Eluent: 10 mM KOH 0–8 min, 10–40 mM 8–14 min, Weigh approximately 30 ± 2 mg of sample on an 40–60 mM 14–20 min, 60 mM 20–30 min Eluent Source: EGC II KOH with CR-ATC analytical balance and then transfer to a previously Flow Rate: 0.40 mL/min weighed 100 mL polypropylene volumetric flask. Temperature: 30 °C Inj. Volume: 100 µL Dissolve the solid in 100 mL of ACS grade MeOH (d Matrix Elim. Vol.: 1 mL Concentrator: UTAC-ULP1 (5 × 23 mm) = 0.7918 g/mL) to prepare a final sample concentration Detection: ASRS 300, 2 mm, recycle mode, 60 mA of 0.30 mg/mL (w/v). Caution: MeOH is flammable. Peaks: 1. Fluoride 20 µg/L 2. Chloride 30 Work under a hood. Record the weight of this solution in 3 3. Nitrite 100 the volumetric flask and subtract from the weight of the 4. Carbonate – 5. Sulfate 150 5 empty volumetric flask and solid to obtain the weight of 6. Bromide 100 7. Nitrate 100 MeOH used to prepare the sample. To completely dissolve 8. Phosphate 150 the solid material, sonicate the solution for approximately

3 15 min. µS 4 1 8 6 7 SYSTEM PREPARATION AND SETUP 2 1. Install an EGC II KOH cartridge in the EG-3000 module. 2. Install backpressure tubing in place of the columns to produce a total backpressure of ~2000-2500 psi at a 0 0 10 20 30 flow rate of 1 mL/min. Minutes 25923 3. Condition the cartridge by setting the KOH

Figure 1. Separation of common inorganic anions in a 1000-fold concentration to 50 mM at 1 mL/min for 30 min. dilution of a mixed common anion standard on the IonPac AS15 4. Disconnect the backpressure tubing installed in place column. of the column set. 5. Install a CR-ATC between the EGC II KOH cartridge and the EGC degas. Primary Dilution Standards 6. Hydrate the CR-ATC prior to use by following Prepare 100 mg/L each of fluoride and phosphate the instructions outlined in the EluGen Cartridge standards in separate 20 mL scintillation vials by Quickstart Guide. combining 2 mL of the respective 1000 mg/L stock 7. Install 2 × 50 mM AG15 and 2 × 250 mm AS15 solutions with 18 mL of deionized water. Prepare columns in the lower compartment of the DC 1 mg/L each of chloride, sulfate, and nitrate standards in using red PEEKTM tubing (0.005” i.d.) between separate 125 mL HDPE bottles by combining 100 µL of connections. the respective 1000 mg/L stock solutions with 99.9 mL of 8. Install a 5 × 23 mm UTAC-ULP1 concentrator in deionized water. place of the sample loop on valve #1 using black PEEK (0.010” i.d.) tubing. Direction of sample Calibration Standards loading should be opposite of analytical flow. Prepare calibration standards in the low-µg/L to mg/L 9. Make sure the pressure is ~2200-2500 psi using the range by adding the appropriate volumes from the target operating conditions described earlier to allow the anions primary dilution standards to separate degas assembly to effectively remove electrolysis 125 mL HDPE bottles and dilute to 100 mL with gases. If necessary, install additional backpressure deionized water. Four levels of calibration standards were tubing or trim tubing between degas assembly and the used in this study to cover the expected concentrations injection valve to achieve the recommended pressure. found in the pharmaceutical sample. 10. Hydrate and install ASRS 300 suppressor and Carbonate Removal Device (CRD 200) according to the instructions in the product manuals. Install both in recycle mode using red PEEK tubing for all connections.

66 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection The AS autosampler was used in this AN to Click the mouse pointer on the next line and then concentrate the sample and eliminate the matrix from select Reagent Flush from the drop down menu. To the UTAC-ULP1 concentrator column. To install and use a vial as the source of the matrix elimination configure the AS autosampler: solution, as described in #1 above, enter the 1. Install a 1-mL sample syringe (P/N 055066). appropriate vial # in the box. To use the second 2. From the front panel (and under System Parameters), option, described in #2 above, choose the appropriate configure the AS autosampler sample mode to Reagent Reservoir that contains the solution used to "Concentrate." eliminate the matrix. For the volume, enter 1000 µL 3. Connect the AS injection port tubing directly to and Valve Position should equal No Change. Click the injection valve. Be sure the tubing is properly insert to insert the line in the sampler options steps. calibrated before operating the autosampler. This completes the steps required to concentrate 4. The AS autosampler Concentrate option allows the the sample and eliminate the matrix from the AS to deliver sample to a low pressure concentrator concentrator column. at a maximum pressure of 100 psi. Therefore, the sample syringe dispense speed should be no greater RESULTS AND DISCUSSION than 2 in the Chromeleon program. A primary consideration in the development of a suitable IC method for pharmaceuticals is the solubility This application requires a matrix elimination of the API in water. Many drugs and intermediates are step using deionized water to remove MeOH from the insoluble in water and other aqueous solutions that are concentrator column. There are two possible procedures typically used in IC systems. This poses a potential to accomplish this task: analytical challenge as it could lead to precipitation of 1. Rinse a 10 mL AS sample vial several times with the API in the chromatography system and therefore deionized water and then fill the vial with deionized cause excess backpressure and column contamination.9 water. When performing the matrix elimination step To overcome this challenge, a sufficient amount of in the program, direct the AS autosampler to aspirate organic solvent can be added to the eluent to maintain 1 mL from the vial. Separate vials are strongly the solubility of the API or the API can be precipitated recommended for different calibration standards and the resulting solution filtered prior to analysis.9 The and samples to minimize cross contamination. former approach requires a manually prepared eluent The deionized water in the vial should be changed and therefore precludes the use of a Reagent-Free frequently. For ease-of-use, this option for performing ion chromatography (RFIC) system, while the latter the matrix elimination step was used. increases analysis complexity that can lead to potential 2. Alternatively, the matrix elimination step can be contamination and measurement errors. performed by using the sample prep syringe of This AN describes the development of an IC method the AS autosampler with a 5 mL syringe installed. for the determination of monovalent to polyvalent However, this setup requires an 8.2-mL sampling inorganic anions commonly found in pharmaceuticals. needle assembly (P/N 061267) to accommodate the The method combines preconcentration with matrix larger volume. The use of the sample prep syringe elimination to detect trace concentrations of inorganic for eliminating the matrix from the concentrator will impurities in a proprietary water-insoluble drug. A 100 require more time per injection. µL of the pharmaceutical dissolved in 100% MeOH is 3. To setup the concentrate and matrix elimination steps concentrated on a UTAC-ULP1 concentrator column in Chromeleon, use the program wizard and go to the to trap the inorganic anion impurities, while the MeOH Sampler Options section. By default, the first line of matrix is eliminated with 1 mL of deionized water before the Sampler Options steps should appear. The first analysis. This approach eliminates the need for organic line should read: solvent in the eluent or the offline precipitation of the API and therefore improves the methods ease-of-use. 1 Concentrate Loadposition Aspirate = 3 Dispense = 1 Change the dispense speed from 1 to 2 and click Enter.

67 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection The IonPac AS15 column was chosen as the To establish a suitable concentration range for the separation column because it is a high-capacity, target anions, the anions detected in the pharmaceutical hydroxide-selective column specifically developed for the sample (0.30 mg/mL) and MeOH matrix blank were rapid and efficient separation of trace concentrations of compared against a 1000-fold dilution of a mixed inorganic anions in matrices with varying ionic strength. common anion standard (Figure 1). Table 1 summarizes The use of an electrolytically generated hydroxide the range of the calibration curves and the linearity eluent for this application produces an exceptionally for each target anion. The results demonstrate that the low background and baseline noise and therefore lower calibration curves for the target anions were linear with detection limits, which enables the detection of inorganic correlation coefficients (r2) greater than 0.997. impurities that are less than 0.001% (w/w) in the 0.30 mg/ Table 1 also summarizes the estimated limits of detections mL pharmaceutical sample analyzed in this study. (LODs) for the target analytes, calculated based on three It is important to establish a matrix blank and times the signal-to-noise ratio (S/N). ensure its stability before proceeding to analyze the sample. In this AN, MeOH was required to dissolve the pharmaceutical sample. In general, organic solvents are Table 1. Calibration Data and Detection Limits known to contain trace concentrations of inorganic anions Analyte Range Linearity Estimated and low molecular weight organic acids as discussed 2 (µg/L) (r ) Limits of in AU163.10 However, trace anions in solvents can Detectiona also be derived from sample handling procedures and (µg/L) contaminated materials used to transport the solution Fluoride 500–2000 0.9996 0.16 for analysis. Therefore, it is critical to use the same set Chloride 10–100 0.9989 0.39 of containers and other components used to prepare Sulfate 5.0–50 0.9974 0.46 the samples to obtain a representative blank. As shown Nitrate 10–100 0.9997 1.3 in Figure 2, trace concentrations of fluoride, chloride, Phosphate 250–1000 0.9997 1.7 sulfate, and nitrate were detected in 100% MeOH. aLODs estimated from 3 × S/N

The method performance was evaluated by analyzing Column: IonPac AG15, AS15, 2 mm Eluent: 10 mM KOH 0–8 min, 10–40 mM 8–14 min, three different preparations of the pharmaceutical sample 40–60 mM 14–20 min, 60 mM 20–30 min over three days. Figure 3 demonstrates the applicability Eluent Source: EGC II KOH with CR-ATC Flow Rate: 0.40 mL/min of the method for determining trace anions in a Temperature: 30 °C 2.5 Inj. Volume: 100 µL 0.30 mg/mL proprietary pharmaceutical product. As Matrix Elim. Vol.: 1 mL shown, the pharmaceutical sample primarily consists of Concentrator: UTAC-ULP1 (5 × 23 mm) Detection: ASRS 300, 2 mm, recycle mode, 60 mA fluoride and phosphate with only trace concentrations Peaks: 1. Fluoride 5.3 µg/L of chloride, sulfate, and nitrate. When the sample is 2. Chloride 27.3 3. Carbonate – corrected for the MeOH blank, the concentrations of the 3 4. Sulfate 3.8 5. Nitrate 31.5 trace anions (chloride, sulfate, nitrate) were determined to µS be significantly less than the background concentrations. 4 Therefore, this AN focuses on the primary anion cons- 2 5 tituents, fluoride and phosphate, in the pharmaceutical 1 sample. The average concentrations for fluoride and phosphate detected in the sample over three days were 967

0.4 ± 12 µg/L and 339 ± 10 µg/L, respectively. The presence 0 10 20 30 Minutes of fluoride in some inorganic raw materials used for the 25924 preparation of pharmaceuticals is well-known. Calcium Figure 2. Target anions detected in a representative MeOH blank. salts are the most contaminated with fluoride due to their manufacturing process. The determination of fluoride

68 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection Column: IonPac AG15, AS15, 2 mm Table 2. Summary of Data Obtained for Target Anions Eluent: 10 mM KOH 0–8 min, 10–40 mM 8–14 min, in a Water-Insoluble Pharmaceutical Product 40–60 mM 14–20 min, 60 mM 20–30 min Eluent Source: EGC II KOH with CR-ATC Day Analyte Amount % (w/w) in a Retention Peak Flow Rate: 0.40 mL/min Found 0.30 mg/mL Time RSDa Area Temperature: 30 °C (µg/L) Pharmaceutical RSDa Inj. Volume: 100 µL Matrix Elim. Vol.: 1 mL 1 Fluoride 973.5 0.25 0.06 0.12 Concentrator: UTAC-ULP1 (5 × 23 mm) Detection: ASRS 300, 2 mm, recycle mode, 60 mA Phosphate 328.9 0.08 0.02 1.1 Sample: 0.30 mg/mL pharmaceutical (Day 3) 2 Fluoride 953.9 0.24 0.12 0.41 Peaks: 1. Fluoride 975 µg/L 3.99 1 2. Chloride 33.3 Phosphate 349.0 0.09 0.01 0.76 3. Carbonate – 4. Sulfate 6.5 3 Fluoride 974.6 0.25 0.04 0.38 5. Nitrate 33.8 6. Phosphate 339 Phosphate 339.0 0.09 0.01 0.40 an = 6

6 µS 3 CONCLUSION 2 4 5 In this AN, we demonstrated the ability to determine trace anions in a proprietary water-insoluble pharmaceutical using preconcentration with matrix 0.18 elimination. This method was designed to provide a 0 10 20 30 Minutes simpler approach that avoids the potential complications 25925 of column contamination and excess column backpressure Figure 3. Determination of inorganic anion impurities in a that can occur when analyzing water-insoluble samples. proprietary water-insoluble pharmaceutical compound The use of a hydroxide-selective AS15 column provided an efficient separation of common anions from low to high µg/L concentrations that are typically found in pharmaceuticals is critical because excess fluoride in pharmaceuticals. In addition, the combination of a is toxic and can cause bone diseases, such as fluorosis, hydroxide-selective column with an electrolytically osteoporosis, and skeletal fragility.11 The presence of generated potassium hydroxide eluent eliminates the phosphate is also not uncommon in pharmaceuticals problems associated with the manual preparation of as phosphate buffers are commonly used during the hydroxide eluents and therefore further increases the ease- preparation of the final formulation. Table 2 summarizes of-use and method automation. This method demonstrated the results for the determination of fluoride and phosphate good linearity, sensitivity, precision, and accuracy for in the pharmaceutical sample. For the three day study, determining inorganic anion impurities in a water- the retention time and peak area RSDs were <0.1% and insoluble pharmaceutical compound. <1.2%, respectively, for the target anions. The method accuracy was also evaluated by determining the recoveries List of Suppliers of fluoride and phosphate spiked into the sample at VWR Scientific, P.O. Box 7900, San Francisco, CA concentrations that were nearly equivalent to the unspiked 94120, USA. Tel: 1-800-252-4752. www.vwr.com sample. The calculated recoveries for fluoride and phosphate were 102.6% and 107.7%, respectively. The good recoveries obtained in this study indicate that the method performed well for the determination of the target anions in a proprietary water-insoluble pharmaceutical compound.

69 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection REFERENCES 8. Determination of Sulfate Counter Ion and Anionic 1. Impurities in New Drug Substances, International Impurities in Aminoglycoside Drug Substances by Conference on Harmonisation of Technical Ion Chromatography with Suppressed Conductivity Requirements for Registration of Pharmaceuticals Detection. Application Note 190 (LPN 1946, for Human Use, ICH Guideline Q3A (R2), 2006. September 2007), Dionex Corporation, available at http://www.ich.org/LOB/media/ Sunnyvale, CA. MEDIA422.pdf 9. Cassidy, S.A.; Demarest, C.W.; Wright, P.B.; 2. Roy, J. Pharmaceutical Impurities – A Mini Review. Zimmerman, B. Development and Application of AAPS Pharm. Sci. Tech. 2002, 3(2), 1–8. a Universal Method for Quantitation of Anionic 3. Hulse, W.L.; Grimsey, I.M.; De Matas, M. The Constituents in Active Pharmaceutical Ingredients Impact of Low-Level Inorganic Impurities on Key During Early Development Using Suppressed Physicochemical Properties of Paracetamol. Int. J. Conductivity Ion Chromatography. J. Pharm. Pharm. 2008, 349, 61–65. Biomed. Anal. 2004, 34, 255–264. 4. Basak, A.K.; Raw, A.S.; Al Hakim, A.H.; Furness, 10. Determination of Trace Anions in Organic Solvents S.; Samaan, N.I.; Gill, D.S.; Patel, H.B.; Powers, Using Matrix Elimination and Preconcentration. R.F.; Yu, L. Pharmaceutical Impurities: Regulatory Application Update 163 (LPN 1962, October 2007), Perspective for Abbreviated New Drug Applications. Dionex Corporation, Sunnyvale, CA. Adv. Drug Deliv. Rev. 2007, 59, 64-72. 11. Bouygues-de Ferran, A.M.; Pham-Huy, C.; Postaire, 5. Ion Chromatography in the Pharmaceutical Industry. M.; Hamon, M. Determination of Trace Amounts of Application Note 106 (LPN 0660, July 1996), Dionex Fluoride in Raw Materials for Pharmaceuticals by Corporation, Sunnyvale, CA. Gas-Liquid Chromatography. J. Chromatogr. 1991, 6. Quantification ofAnions in Pharmaceuticals. 585, 289–295. Application Note 116 (LPN 0924-01, June 2004), Dionex Corporation, Sunnyvale, CA. 7. Assay for Citrate and Phosphate in Pharmaceutical Formulations Using Ion Chromatography. Application Note 164 (LPN 1643, August 2004), Dionex Corporation, Sunnyvale, CA.

70 Determination of Inorganic Anion Impurities in a Water-Insoluble Pharmaceutical by Ion Chromatography with Suppressed Conductivity Detection Application Note 117

Quantification of Carbohydrates and Glycols in Pharmaceuticals

INTRODUCTION the analysis of carbohydrate and glycol ingredients in The United States Food and Drug Administra­ pharmaceutical formulations are compared. The ICE- tion (U.S. FDA)1–3 and the regulatory agencies in other AS1 resin bead is a completely sulfonated polystyrene/ countries require that pharmaceutical products be tested divinylbenzene polymer with a capacity of about for composition to verify their identity, strength, quality, 27 meq/column and with moderate hydrophilic char­ and purity. Recently, attention has been given to inactive acteristics. The retention mechanisms possible in this ingredients as well as active ingredients. Some of these column include ion exclusion, steric exclusion, and ingredients are nonchromophoric and cannot be detected adsorption. Weakly-ionized acids are separated by pKa by absorbance. Some nonchromo­phoric ingredients, differences, size, and hydrophobicity. This column is such as carbohydrates, glycols, sugar alcohols, amines, ideal for the determination of aliphatic organic acids and and sulfur-containing compounds, can be oxidized, and alcohols in complex or high-ionic strength samples. therefore can be detected using amperometric detec­ The CarboPac PA10 column packing consists of tion. This detection method is specific for those analytes a nonporous, highly crosslinked polystyrene/divinyl­ that can be oxidized at the selected potential, leaving all benzene substrate agglomerated with 460‑nm diameter other nonoxidizable compounds transparent.4–5 Ampero­ latex. The MicroBead™ latex is functionalized with metric detection is a powerful detection technique with a quaternary ammonium ions, which create a thin surface broad linear range and very low detection limits. rich in anion-exchange sites. The packing is specifically This Application Note describes the use of three designed to have a high selectivity for monosaccharides. different anion-exchange columns with amperometric The PA10 has an anion-exchange capacity of approxi­ detection to analyze common simple sugars, sugar alco­ mately 100 µeq/column. hols, and glycols in pharmaceutical formulations. Two The CarboPac MA1 resin is composed of a poly­ oral, over-the-counter medications were selected styrene/divinylbenzene polymeric core. The surface is as representative pharmaceutical products. A cough sup­ grafted with quaternary ammonium anion-exchange pressant and a multisymptom cold/flu medication were functional groups. Its macroporous structure provides chosen because they contain a complex mixture an extremely high anion-exchange capacity of 1450 of simple sugars, glycols, and sugar alcohols. These µeq/column. The CarboPac MA1 column is designed carbohydrates and glycols are also commonly found specifically for sugar alcohol and glycol separations. The in other medications. Furthermore, these formulations ICE-AS1 and PA10 columns, but not the MA1 column, contain inorganic and organic anionic ingredients that are compatible with eluents containing organic solvents, have been analyzed using the IonPac® AS14 and AS11 which can be used to clean these columns. anion-exchange columns with suppressed conductivity Expected detection limits, linearity, selectivity, ac­ detection.6 curacy, and precision are reported for the CarboPac MA1 In the methods outlined in this Note, the selectivi­ column. The performance of the CarboPac PA10 column ties of the IonPac ICE-AS1 ion exclusion, CarboPac™ for monosaccharide analysis is presented in Technical PA10, and CarboPac MA1 anion-exchange columns for Note 40.7

71 Quantification of Carbohydrates and Glycols in Pharmaceuticals EQUIPMENT Mannitol, ACS grade (J. T. Baker Incorporated) Dionex DX-500 system consisting of: Maltitol (Aldrich Chemical Co.) GP40 Gradient Pump, with degas option Glucose, reference grade (Pfanstiehl Laboratories) ED40 Electrochemical Detector Sucrose, ACS certified (Fisher Scientific) LC30 or LC25 Chromatography Oven AS3500 Autosampler On-line degassing is necessary because the am­ perometric detector is sensitive to oxygen in the eluent. PeakNet Chromatography Workstation High flow rates of 2 mL/minute, such as those used in the ICE-AS1 system, require more frequent on-line REAGENTS AND STANDARDS degassing intervals (every 5 minutes) to eliminate cyclic Reagents baseline drift. Sodium hydroxide, 50% (w/w) (Fisher Scientific) Perchloric acid, 70% (w/w) (Fisher Scientific) PREPARATION OF SOLUTIONS AND REAGENTS Deionized water, 18 MΩ-cm resistance or higher Perchloric Acid Eluent 100 mM Perchloric acid Standards Degas 1983 mL of deionized water for 20 minutes Propylene glycol, anhydrous (Sigma Chemical Co.) and mix with 17.1 mL 70% (w/w) perchloric acid. Con­ Glycerol (EM Science) nect the eluent reservoir to the instrument and pressurize Sorbitol (Eastman Chemical Company) with helium.

Conditions System 1 System 2 System 3 Columns: IonPac ICE-AS1 Analytical CarboPac PA10 Analytical CarboPac MA1 Analytical (P/N 43197) (P/N 46110) (P/N 44066) NG1 Neutral Guard CarboPac PA10 Guard CarboPac MA1 Guard (P/N 39567) (P/N 46115) (P/N 44067) Flow Rate: 2.0 mL/min 1.5 mL/min 0.4 mL/min Injection Volume: 10 µL 10 µL 10 µL Oven Temperature: 30 °C 30 °C 30 °C Detection (ED40): Integrated amperometry, Integrated amperometry, Integrated amperometry, platinum electrode gold electrode gold electrode Waveform for ED40: Time Potential Integration Time Potential Integration Time Potential Integration (s) (V) (Begin/End) (s) (V) (Begin/End) (s) (V) (Begin/End) 0.00 +0.30 0.00 +0.05 0.00 +0.05 0.05 +0.30 Begin 0.20 +0.05 Begin 0.20 +0.05 Begin 0.25 +0.30 End 0.40 +0.05 End 0.40 +0.05 End 0.26 +1.40 0.41 +0.75 0.41 +0.75 0.60 +1.40 0.60 +0.75 0.60 +0.75 0.61 +0.10 0.61 -0.15 0.61 -0.15 1.00 +0.10 1.00 -0.15 1.00 -0.15 Eluent Components: A: 100 mM Perchloric acid A: Water A: Water B: 200 mM Sodium hydroxide B: 1.0 M Sodium hydroxide Eluent Concentration: 100 mM Perchloric Acid 18 mM Sodium hydroxide 480 mM Sodium hydroxide Method: Time A (%) Time A (%) B (%) Time A (%) B (%) (min) (min) (min) 0.0 100 0.0 91 9 0.0 52 48 End 100 11.0 91 9 60.0 52 48 11.1 0 100 17.6 0 100 17.7 91 9 40.0 91 9

72 Quantification of Carbohydrates and Glycols in Pharmaceuticals Sodium Hydroxide Eluents SAMPLE PREPARATION 200 mM Sodium hydroxide Dilute viscous products with water on a weight per It is essential to use high-quality water of high resis­ weight (w/w) basis. Combine 1 gram of medication with tivity (18 MΩ-cm) as free of dissolved carbon dioxide 9 grams of water to obtain a 10-fold dilution. Further as possible. Biological contamination should be absent. dilute the medication to yield 100 and 1000‑fold dilu­ Additionally, borate, a water contaminant that can break tions on a weight per weight (w/w) basis. Determine through water purification cartridges (prior to any other product densities by measuring the weights of known indication of cartridge depletion), can be removed by volumes. Calculate the final concentrations based on placing a BorateTrap column (P/N 47078) between the the densities of these medications. The ingredients of pump and the injection valve. It is extremely important each medication are presented in Tables 1 and 2. The to minimize contamination by carbonate, a divalent ingredients noted in bold-face type can be analyzed by anion at high pH that binds strongly to the columns, anion-exchange chromatography with amperometric de­ causing a loss of chromatographic resolution and ef­ tection. Many of the other ingredients listed below can ficiency. Commercially available sodium hydroxide be analyzed using the IonPac AS14 and AS11 columns pellets are covered with a thin layer of sodium carbonate with suppressed conductivity detection.6 and should not be used. A 50% (w/w) sodium hydroxide Any purified water used for dilutions should be solution is much lower in carbonate and is the preferred tested for trace carbohydrates prior to its use. Test the source for sodium hydroxide. sample containers for residual carbohydrates prior to Dilute 20.8 mL of a 50% (w/w) sodium hydroxide use by adding pure water, shaking or vortexing, and solution into 1980 mL of thoroughly degassed water to then testing the liquid. Prerinsing the vials with purified yield a 200 mM sodium hydroxide solution. Keep the water can eliminate artifacts and erroneous results. eluents blanketed under 5–8 psi (34–55 kPa) of helium at all times. DISCUSSION AND RESULTS Selectivity 1.0 M Sodium hydroxide Figure 1 shows the separation of sorbitol, glycerol, Follow the same precautions described above for and propylene glycol standards using a 100 mM per­ the 200 mM sodium hydroxide eluent. Dilute 104 mL chloric acid eluent with the IonPac ICE-AS1 analytical of a 50% (w/w) sodium hydroxide solution into 1896 column and the NG1 guard column. The separation mL of thoroughly degassed water to yield 1.0 M sodium was isocratic, which decreases injection-to-injection hydroxide. Keep the eluents blanketed under 5–8 psi run times and increases sample throughput. Glycerol, (34–55 kPa) of helium at all times. propylene glycol, and sugar alcohols were determined within 10 minutes. Sucrose, maltitol, and mannitol were STOCK STANDARDS not well resolved by this method (results not shown). Solid standards were dissolved in purified water to Oxygen is reduced using the same waveform used to 10 g/L concentrations. These were combined and fur­ ther diluted with purified water to yield the desired stock mixture concentrations. Table 1. Cough suppressant ingredients For determinations of linear range, combine 10 g/L solutions of propylene glycol, glycerol, sorbitol, manni­ Type tol, glucose, maltitol, and sucrose to make a 1.0 g/L stan­ Dextromethorphan Hydrobromide Active dard mix solution. Dilute with water to concentrations of Citric Acid Inactive 800, 600, 400, 200, 100, 80, 60, 40, 20, 10, 8, 6, 4, 2, 1, FD&C Red 40 Inactive 0.8, 0.6, 0.4, 0.2, and 0.1 mg/L. Maintain the solutions in Flavors Inactive a frozen state at –20 °C until needed. Glycerin (glycerol) Inactive Propylene Glycol Inactive Saccharin Sodium Inactive Sodium Benzoate Inactive Sorbitol Inactive Water Inactive

73 Quantification of Carbohydrates and Glycols in Pharmaceuticals suppressant. Figure 5B is an expanded view of Figure Table 2. Multisymptom Cold/Flu Ingredients 5A to better reveal the minor peaks. In general, peak Type elution order from the ICE-AS1 column is the reverse Pseudoephedrine Hydrochloride Active of that of the MA1 column. The early eluting peaks for Acetaminophen Active sugar alcohols and carbohydrates were not resolved Dextromethorphan Hydrobromide Active on the ICE-AS1 column. Propylene glycol and glyc­ Citric Acid Inactive erol eluted later and were completely resolved on this FD&C Yellow #6 Inactive Flavor Inactive column. Figure 6 shows the analysis of a multisymptom Glycerin (glycerol) Inactive cold/flu medication using the MA1 column. Polyethylene Glycol Inactive Propylene Glycol Inactive Purified Water Inactive Column: IonPac ICE-AS1, NG1 Guard Saccharin Sodium Inactive Eluent: 100 mM Perchloric acid Sodium Citrate Inactive Flow Rate: 2.0 mL/min Sucrose Inactive Inj. Volume: 10 µL Detection: Integrated amperometry, platinum electrode 500 Peaks: 1. Sucrose 1 mg/L detect sugar alcohols. Oxygen dissolved in the sample 2. Sorbitol 1 3 3. Glycerol 1 eluted at the total permeation volume, and appeared as 4. Propylene glycol 1 a dip in the baseline at just before 8 minutes. Use of the 2 NG1 neutral guard column increased the total perme­ nC ation volume and moved the oxygen dip away from the 4

analytes. Without the NG1, sucrose, sorbitol, maltitol, 1 and mannitol were slightly better resolved, but the 0 oxygen dip encroached on the propylene glycol peak. An alternative way to reduce or eliminate the oxygen dip is to degas the sample prior to injection. 02468 10 Minutes The CarboPac PA10 column separated propylene 13154 glycol, glycerol, sorbitol, mannitol, maltitol, glucose, Figure 1. Separation of common glycols, sugar alcohols, and and sucrose using an isocratic sodium hydroxide eluent carbohydrates in pharmaceutical formulations on an IonPac ICE- (Figure 2). Propylene glycol and glycerol eluted near the AS1 column. void and were not baseline-resolved. Propylene glycol, glycerol, sorbitol, mannitol, maltitol, and sucrose were completely resolved when Column: CarboPac PA10 and Guard analyzed on the CarboPac MA1 (Figure 3). Shorter run Eluent: 18 mM Sodium hydroxide times are possible for both the PA10 and MA1 methods Flow Rate: 1.5 mL/min Inj. Volume: 10 µL 120 4 by adjusting the eluent strength, but resolution may 3 Detection: Integrated amperometry, be lost. gold electrode Peaks: 1. Propylene glycol Although not presented here, nearly a dozen over- 2. Glycerol 3. Sorbitol the-counter medications have been analyzed using the 4. Mannitol 2 ICE-AS1 and MA1 columns with amperometric detection. nC 5. Maltitol 6. Glucose These medications include both solid and liquid formu­ 5 7. Sucrose 1 6 lations such as nasal and oral decongestants, astringents, 7 antacids, enemas, sleep aids, analgesics, cleaning and disinfecting solutions, antihistamines, and allergy syr­ 0 ups. In most cases, the known carbohydrate ingredients 0 2 468 10 12 14 in each formulation were separated from each other us­ Minutes ing the MA1 column without any apparent interference. 13155 Figures 4 and 5 compare the selectivity of the Figure 2. Separation of common glycols, sugar alcohols, and ICE‑AS1 and MA1 columns for the analysis of a cough carbohydrates in pharmaceutical formulations on a CarboPac PA10 column.

74 Quantification of Carbohydrates and Glycols in Pharmaceuticals Column: CarboPac MA1, MA1 guard Column: CarboPac MA1, MA1 guard Peaks: 1. Propylene glycol 1.1 g/L Eluent: 480 mM Sodium hydroxide Eluent: 480 mM Sodium hydroxide 2. Glycerol 0.6 3. Unidentified — Flow Rate: 0.4 mL/min Flow Rate: 0.4 mL/min 4. Unidentified — 90 Inj. Volume: 10 µL Inj. Volume: 10 µL 5. Unidentified — Detection: Integrated amperometry, Detection: Integrated amperometry, 6. Sorbitol 1.5 2 gold electrode gold electrode 7. Mannitol 0.1 Peaks: 1. Propylene glycol 10 mg/L Sample: 100‑fold dilution (w/w) 8. Unidentified — 2. Glycerol 10 9. Unidentified — 3. Sorbitol 10 10. Unidentified — 11. Maltitol 0.005 4. Mannitol 10 A: 0–6000 nC 5. Maltitol 10 12. Unidentified — nC 6. Sucrose 10 3 4 6000 6

2 5 nC

1 6 1 3 5 4 7 8 9 10 11 12 0 0

0510 15 20 25 30 35 40 01020304050 Minutes 12759 B: Expanded view, 0–400 nC

1 2 6 Figure 3. Separation of glycols, sugar alcohols, and carbohy- 400 drates using the CarboPac MA1 column. 7

nC

3 5 4 8 9 10 11 12 Column: IonPac ICE-AS1, NG1 Guard 0 Eluent: 100 mM Perchloric acid Flow Rate: 2.0 mL/min 0510 15 20 25 30 35 40 Inj. Volume: 10 µL Minutes Detection: Integrated amperometry, 12762 platinum electrode Sample: 100-fold dilution (w/w) Figure 5. Separation of sugar alcohols and glycols in cough sup- 800 Peaks: 1. Mannitol pressant using the CarboPac MA1 column. 2. Sorbitol 3. Glycerol 3 4. Propylene glycol 2 Column: CarboPac MA1, MA1 guard nC 4 Eluent: 480 mM Sodium hydroxide Flow Rate: 0.4 mL/min Inj. Volume: 10 µL 1 2500 Detection: Integrated amperometry, gold electrode 2 Sample: 100-fold dilution (w/w) 0 Peaks: 1. Propylene glycol 1.3 g/L 2. Glycerol 0.3 3. Unidentified — 011234567 98 0 4. Sucrose 1.5 Minutes nC 13156 1

4 Figure 4. Separation of sugar alcohols and glycols in cough suppressant using the IonPac ICE-AS1 column.

3 0

01020304050 Minutes 12764

Figure 6. Sugar alcohols, glycols, and carbohydrates in multi- symptom cold/flu medication.

75 Quantification of Carbohydrates and Glycols in Pharmaceuticals The selection of the best method depends on the ana­ Table 3. Estimated Detection Limits lytes tested. If a rapid analysis of sugar alcohols and simple Using the CarboPac MA1 sugars is needed, with little interest in propylene glycol or glycerol, the CarboPac PA10 column is the best choice. If a ng µg/L rapid analysis of glycerol and propylene glycol is desired, Propylene glycol 4 400 with little interest in carbohydrates, the IonPac ICE-AS1 is Glycerol 0.7 70 the best choice. If the analysis of all of these compounds is Sorbitol 1 100 required, a CarboPac MA1 is the best choice. Mannitol 1 100 Glucose 1 100 Method Detection Limits Maltitol 2 200 Sucrose 7 700 The method detection limits (MDL) for a 10-µL in­ jection of common pharmaceutical constituents using the MA1 column are shown in Table 3. The MDL is defined as the minimum concentration required to produce a sig­ Table 4. Peak Area and Retention Time Precision nal-to-noise ratio of 3. The MDL can be further decreased by increasing the injection volume above the 10-µL injec­ RSD (%) tion volume used in this Application Note, and by using Area Retention Time (min) smoothing algorithms available in PeakNet software.7 Propylene glycol 2.4 0.0 Glycerol 3.0 0.2 Linearity Sorbitol 2.9 0.1 Propylene glycol, glycerol, sorbitol, mannitol, glu­ Mannitol 2.9 0.1 Glucose 2.7 0.1 cose, maltitol, and sucrose standards ranging from 0.1 Maltitol 2.6 0.1 to 1000 mg/L (1 ng to 10,000 ng) were injected (n = 2 Sucrose 3.7 0.1 to 3 per concentration) onto a CarboPac MA1 column. The method was found to be linear for propylene glycol, sorbitol, mannitol, glucose, maltitol, and sucrose over this range (r2 ≥ 0.999). Glycerol was linear over the range Table 5. Recovery of Carbohydrates in Medications of 0.1 to 200 mg/L (1 ng to 2,000 ng per injection; r2 = Percent Recovery 0.999). For the range of 0.1 to 1000 mg/L (1 ng to 10,000 2 Cough Multisymptom ng), glycerol deviated from linearity (r = 0.995). Using Analyte Suppressant Cold/Flu Medication a second order polynomial regression, the r2 for glycerol Propylene glycol 96 111 was 0.9998 over the range of 0.1 to 1000 mg/L (1 ng to Glycerol 98 109 10,000 ng). For all analytes, linearity was demonstrated Sorbitol 87 105 over at least three orders of magnitude. Broad linear Mannitol 99 105 ranges help eliminate the need to repeat sample analyses Glucose 101 104 when components vary greatly in concentration. Maltitol 103 106 Calibration curves for the MA1 column are Sucrose 103 114 presented in Figure 7.

Precision Recovery from Sample Matrix The peak area and retention time RSDs for 10-mg/L To assess the accuracy of this method, evaluate both injections of standards (10 µL per injection, 12 injec­ medications by the method of standard addition. Com­ tions) run on the MA1 are presented in Table 4. RSDs bine each 1000-fold (w/w) diluted formulation with an varied from 2–4% at this concentration. Precision is equal weight of a 100 mg/L mixture of propylene glycol, affected by concentration; RSD values increase as the glycerol, sorbitol, mannitol, maltitol, sucrose, and concentrations approach the MDL. RSDs increase near glucose to yield a 50 mg/L spiked solution [2000-fold the MDL because peak integration becomes less precise dilution (w/w)]. Subtract the amount of each analyte from the contribution of variation in baseline noise from measured in the sample before it was spiked from the run-to-run. total amount of each analyte measured in the spiked

76 Quantification of Carbohydrates and Glycols in Pharmaceuticals sample to yield the amount of spiked analyte recovered. Concentration of Known Ingredients in Pharmaceutical The amount of spiked analyte recovered relative to the Products known amount added then yields the percent recovery. Ingredients in pharmaceutical products that could be Figure 5 shows the separation of carbohydrate and identified by retention times are listed in Table 6. Their glycol ingredients in cough suppressant using the MA1 respective concentrations were determined and also pre­ column. Figure 6 shows the separation of carbohydrates sented in Table 6. The ingredients listed on the product and glycols in a multisymptom cold/flu formulation. The container are marked with an asterisk (*). It is not the percent recovery after standard addition [50 mg/L spike, normal practice of drug manufacturers to state the con­ 2000-fold dilution, (w/w)] is presented in Table 5. Per­ centrations of inactive ingredients on their product labels; cent recovery ranged from 87 to 114% for the glycols, therefore, the accuracy of these formulations against the sugar alcohols, and carbohydrates tested. stated label concentrations could not be evaluated. In addition to the labeled content of the pharmaceu­ tical products, other carbohydrate or glycol ingredients Table 6 Concentration of carbohydrates may be present. Unlabeled ingredients are not marked with an asterisk. Trace levels of unlabeled carbohydrates in medications were determined by injecting less dilute samples [1000 Concentration (g/L) and 100-fold dilutions (w/w)]. Expanding the chromato­ Cough Multisymptom gram for the cough suppressant (Figure 5B) reveals the Analyte Suppressant Cold/Flu Medication presence of minor peaks. Besides sorbitol, which is a la­ Propylene glycol 184* 215* beled ingredient of the cough suppressant, the unlabeled Glycerol 144* 56* ingredients, mannitol and maltitol, were also identified Sorbitol 288* 0.1 based on their retention time. Mannitol and maltitol are Mannitol 13 1.5 probably trace impurities of sorbitol. Seven other peaks Glucose 0.2 0.2 were detected in the cough suppressant (Figure 5B), but Maltitol 0.2 Not Detected Sucrose Not Detected 459* were not identified. The multisymptom cold/flu medica­ tion contained one unidentified minor peak (Figure 6). * Ingredients that are listed on the product containers.

1200000000

1000000000

2 800000000 Propylene Glycol r = 0.9997 2 Glycerol...... r = 0.9998*

2 Sorbitol...... r = 0.9994

2 600000000 Mannitol...... r = 0.9996

Area Units 2 Glucose...... r = 0.9998

2 Maltitol...... r = 0.9995 400000000 2 Sucrose...... r = 0.9995 * Second order polynomial, glycerol is linear from 200000000 0.1 to 2000 ng/L (r 2 = 0.999)

0 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 ng Injected 13157

Figure 7. Method linearity for CarboPac MA1 with amperometric detection.

77 Quantification of Carbohydrates and Glycols in Pharmaceuticals CONCLUSION REFERENCES Pharmaceutical formulations can be analyzed for 1. CFR Title 21, Food and Drugs, Chapter 1, FDA, B both glycols and carbohydrates using the CarboPac Part 211.22, “Responsibilities of quality control unit.” MA1 column with amperometric detection. The MA1 2. CFR Title 21, Food and Drugs, Chapter 1, FDA, I resolves glycols such as propylene glycol and glycerol. Part 211.160, “General requirements.” It also resolves sugar alcohols such as sorbitol, mannitol, 3. CFR Title 21, Food and Drugs, Chapter 1, FDA, I maltitol, and carbohydrates such as sucrose and glucose 211.165, “Testing and release for distribution.” in the same injection. The CarboPac PA10 also resolves 4. Rocklin, R. A Practical Guide to HPLC Detection; sugar alcohols and carbohydrates, but is not suitable D. Parriott, Ed.; Chapter 6, Electrochemical Detection, for glycols such as propylene glycol and glycerol. The Academic Press: San Diego, CA, 1993, pp 145–173. IonPac ICE-AS1 with the NG1 can be used to rapidly 5. Rocklin, R.D. J. Chromatogr. 1991, 546, 175–187. separate propylene glycol and glycerol, but is less effec­ 6. Dionex Corporation, “Quantification of Anions in tive for many carbohydrates and some sugar alcohols. Pharmaceuticals”, Application Note 116. The ICE-AS1 reverses the elution order of glycols and 7. Dionex Corporation, “Glycoprotein Monosaccharide carbohydrates when compared to the CarboPac columns. Analysis Using High-Performance Anion-Exchange All of these columns use isocratic eluents, which Chromatography with Pulsed Amperometric Detection simplify analysis. The MA1 is a high-capacity column (HPAE-PAD)”, Technical Note 40. and generally has longer run times than the PA10 and the ICE-AS1. Amperometric detection eliminates potential interferences from the nonoxidizable ingredients in the LIST OF SUPPLIERS formulation and provides a sensitive means to detect Fisher Scientific, 711 Forbes Ave., Pittsburgh, nonchromophoric analytes. Carbohydrates, glycols, and Pennsylvania, 15219-4785, U.S.A., 1-800-766-7000. sugar alcohols can be detected at the 70–700 µg/L levels Aldrich Chemical Company, Inc., 1001 West using the CarboPac MA1 column with amperometric Saint Paul Avenue, P.O. Box 355, Milwaukee, detection. The three classes of compounds tested were Wisconsin, 53233, U.S.A., 1-800-558-9160. linear over more than three orders of magnitude using Sigma Chemical Company, P.O. Box 14508, St. Louis, this system. The recoveries from pharmaceutical for­ Missouri, 63178, U.S.A., 1-800-325-3010. mulations were greater than 87% based on the method EM Science, P.O. Box 70, 480 Democrat Road, of standard addition. This method can also be used to Gibbstown, New Jersey, 08027, U.S.A., evaluate trace levels of carbohydrate, glycol, and sugar 1‑800‑222‑0342. alcohol contaminants. Eastman Chemical Company, 1001 Lee Road, Rochester, New York, 14652-3512, U.S.A., 1‑800‑225‑5352. J. T. Baker Incorporated, 222 Red School Lane, Phillipsburg, New Jersey, 08865, U.S.A., 1‑800‑582‑2537. Pfanstiehl Laboratories, Inc., 1219 Glen Rock Avenue, Waukegan, Illinois, 60085-0439, U.S.A. 1‑800-383-0126.

78 Quantification of Carbohydrates and Glycols in Pharmaceuticals Application Note 122

The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths

INTRODUCTION with little interference from other broth ingredients.1–3 Fermentation broths are used in the manufacture Although biosensor and flow-injection analyzer-based of biotherapeutics and many other biological materials methods are commonly used to evaluate fermentation produced using recombinant genetic technology, as well broths, these techniques cannot simultaneously deter­ as for the production of methanol and ethanol as alterna­ mine multiple compounds.4,5 Refractive index detection tive energy sources to fossil fuels. has been used for the analysis of fermentation broths, but Recently, attention has been given to characterizing is limited by poor sensitivity and selectivity.6,7 Postcol­ the ingredients of fermentation broths because carbon umn derivatization with UV/Vis detection has also been sources and metabolic by-products have been found to used, but is complicated by the additional reaction chem­ impact the yield of the desired products. Carbohydrates istry and poor sensitivity.8,9 HPAE-PAD provides the (glucose, lactose, sucrose, maltose, etc.) are carbon analytical capability to monitor, without derivatization, sources essential for cell growth and product synthesis, a large number of different compounds simultaneously while alcohols (ethanol, methanol, sugar alcohols, etc.), using a single instrument and chromatographic method. glycols (glycerol), and organic anions (acetate, lactate, This Application Note describes the use of two formate, etc.) are metabolic by-products that often different anion-exchange columns with amperomet­ reduce yields. ric detection to analyze simple sugars, sugar alcohols, Fermentation broths are complex mixtures of alcohols, and glycols in yeast and bacterial fermentation nutrients, waste products, cells, cell debris, and desired broths. The yeast Saccharomyces cerevisiae in Yeast products. Many of these ingredients are nonchromophor­ Extract-Peptone-Dextrose (YPD) broth and the bacteria ic and cannot be detected by absorbance. Carbohydrates, Escherichia coli (E. coli) in Luria-Bertani (LB) broth are glycols, alcohols, amines, and sulfur-containing com­ common eukaryotic and prokaryotic fermentation sys­ pounds can be oxidized and detected by amperometry. tems, respectively. Both fermentation broth cultures are This detection method is specific for analytes that can be complex and contain undefined media ingredients, and oxidized at a selected potential, leaving all other com­ thus are a great challenge for most separation and detec­ pounds undetected. tion technologies. These formulations contain inorganic Pulsed amperometric detection (PAD) is a powerful and organic anionic ingredients that have been analyzed detection technique with a broad linear range and very using the IonPac® AS11 and AS11-HC anion-exchange low detection limits. High-performance anion-exchange columns with suppressed conductivity detection.10 chromatography (HPAE) is capable of separating com­ In the methods outlined in this Note, the selectivities plex mixtures of carbohydrates. For complex samples of the CarboPac™ PA1 and CarboPac MA1 anion- such as fermentation broths, the high resolving power exchange columns are compared for the analysis of of HPAE and the specificity of PAD allow the determi­ carbohydrate, alcohol, and glycol ingredients in fermen­ nation of carbohydrates, glycols, sugar alcohols (aldi­ tation broths. tols), and other alcohols such as ethanol and methanol,

79 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths The CarboPac PA1 column packing consists of a 2-Deoxy-D-glucose, reference grade (Pfanstiehl 10-µm nonporous, highly crosslinked polystyrene/divi­ Laboratories) nylbenzene substrate agglomerated with 350-nm diam­ Erythritol (Pfanstiehl Laboratories) eter latex. The MicroBead™ latex is functionalized with Ethanol (EM Science) quaternary ammonium ions, which create a thin surface D-Fructose, reference grade (Pfanstiehl Laboratories) rich in anion-exchange sites. The PA1 has a unique Fucose, reference grade (Pfanstiehl Laboratories) MicroBead pellicular resin structure that gives it stability Galactitol, reference grade (Pfanstiehl Laboratories) from pH 0–14 at all concentrations of buffer salts, and D-Galactose, reference grade (Pfanstiehl Laboratories) enables excellent mass transfer, resulting in rapid gradi­ Galactosamine, reference grade (Pfanstiehl ent equilibration. The PA1 has an anion-exchange capac­ Laboratories) ity of approximately 100 µeq/column and is specifically D-Glucosamine, reference grade (Pfanstiehl designed as a general purpose carbohydrate column. Laboratories) The CarboPac MA1 resin is composed of a poly­ β-D-Glucose, reference grade (Pfanstiehl styrene/divinylbenzene polymeric core. The surface is Laboratories) grafted with quaternary ammonium anion-exchange Glycerol (EM Science) functional groups. Its macroporous structure provides an α-Lactose, monohydrate (Sigma Chemical Co.) extremely high anion-exchange capacity of 1450 µeq/ Maltose, monohydrate, reference grade (Pfanstiehl column. The CarboPac MA1 column is designed specifi­ Laboratories) cally for sugar alcohol and glycol separations. Maltitol (Aldrich Chemical Co.) Expected detection limits, linearity, selectivity, and Maltotriose, hydrate (Aldrich Chemical Co.) precision are reported for the CarboPac MA1 column Mannitol, ACS grade (J.T. Baker Inc.) using a Dionex DX‑500 BioLC® system with pulsed Methanol (EM Science) amperometric detection. Raffinose, pentahydrate, reference grade (Pfanstiehl Laboratories) EQUIPMENT L-Rhamnose, monohydrate (Pfanstiehl Laboratories) Dionex DX-500 BioLC system consisting of: D-Ribose, reference grade (Pfanstiehl Laboratories) Ribitol, reference grade (Pfanstiehl Laboratories) GP40 Gradient Pump with degas option Sorbitol (Eastman Chemical Co.) ED40 Electrochemical Detector Sucrose (Fisher Scientific) LC30 or LC25 Chromatography Oven α-α-Trehalose, dihydrate, reference grade (Pfanstiehl AS3500 Autosampler Laboratories) PeakNet Chromatography Workstation D-Xylose, anhydrous (Sigma Chemical Co.)

REAGENTS AND STANDARDS Culture and Media Reagents Bacto YPD Broth (DIFCO Laboratories, Sodium hydroxide, 50% (w/w) (Fisher Scientific and Cat. No. 0428-17-5) J.T. Baker) Bacto Yeast Extract (DIFCO Laboratories, Deionized water, 18 MΩ-cm resistance or higher Cat. No. 0127-15-1) Bacto Peptone (DIFCO Laboratories, Cat. No. 0118- Standards 15-2) D-Arabinose, anhydrous (Sigma Chemical Co.) LB Broth (DIFCO Laboratories, Cat. No. 0446-17-3) L-Arabitol (Aldrich Chemical Co.) Yeast, Saccharomyces cerevisiae, Bakers Yeast type II 2,3-Butanediol (Sigma Chemical Co.) (Sigma Chemical Co., Cat. No. 45C-2) D-Cellobiose, anhydrous (Sigma Chemical Co.) Bacteria, Escherichia coli (donated by SRI International)

80 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths CONDITIONS PREPARATION OF SOLUTIONS AND REAGENTS CarboPac MA1 Method: Sodium Hydroxide Eluents Columns: CarboPac MA1 Analytical (P/N 44066) 100 mM Sodium hydroxide CarboPac MA1 Guard (P/N 44067) It is essential to use high-quality water of high resis­ Flow Rates: 0.4 mL/min tivity (18 MΩ-cm) as free of dissolved carbon dioxide as possible. Biological contamination should be absent. Eluent: A: 480 mM Sodium hydroxide Additionally, borate, a water contaminant that can break Program: Time (min) A (%) through water purification cartridges (prior to any other 0.0 100 indication of cartridge depletion), can be removed by 70.0 (End) 100 placing the BorateTrap™ column (P/N 47078) between the pump and the injection valve. It is extremely impor­ CarboPac PA1 Method: tant to minimize contamination by carbonate, a divalent Columns: CarboPac PA1 Analytical (P/N 35391) anion at high pH that binds strongly to the columns, CarboPac PA1 Guard (P/N 43096) causing a loss of chromatographic resolution and ef­ ficiency. Commercially available sodium hydroxide Flow Rate: 1.0 mL/min pellets are covered with a thin layer of sodium carbonate Eluents: A: Water and should not be used. A 50% (w/w) sodium hydroxide B: 100 mM Sodium hydroxide solution is much lower in carbonate and is the preferred C: 250 mM Sodium hydroxide source for sodium hydroxide. Program: Time (min) A (%) B (%) C (%) Dilute 10.4 mL of 50% (w/w) sodium hydroxide 0.00 84 16 0 solution into 1990 mL of water to yield 100 mM sodium 60.0 84 16 0 hydroxide. Keep the eluents blanketed under 34–55 kPa (5–8 psi) of helium at all times. 60.1 0 0 100 70.0 0 0 100 250 and 480 mM Sodium hydroxide 70.1 84 16 0 When preparing these eluents, follow the same 90.0 (End) 84 16 0 precautions described above for the 100 mM sodium hydroxide eluent. Common to Both Methods: Injection Volume: 10 µL Sodium 50% Sodium Water Hydroxide (mM) Hydroxide (mL) (mL) Temperature: 30 ºC 250 26 1974 Detection (ED40): Pulsed amperometry, 480 50 1950 gold electrode Waveform for the ED40: Keep the eluents blanketed under 34–55 kPa (5–8 Time (s Potential (V) Integration psi) of helium at all times. On-line degassing is neces­ 0.00 +0.05 sary because the amperometric detector is sensitive to 0.20 +0.05 Begin oxygen in the eluent. Set the pump to degas for 0.40 +0.05 End 30 seconds every 4 minutes. 0.41 +0.75 0.60 +0.75 STOCK STANDARDS Keep solid standards desiccated and under vacuum 0.61 ­0.15 prior to use. Dissolve in purified water to 10 g/L con­ 1.00 ­0.15 centrations. Combine and further dilute with purified water to yield the desired stock mixture concentrations. Maintain the solutions frozen at –20 °C until needed.

81 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths For determinations of linear range and lower detec­ ribose, arabinose, rhamnose, and raffinose) in 200 mL of tion limits, combine 10 g/L solutions of 2,3-butanediol, purified sterile water. Dissolve 1.0 g of yeast (Saccharo­ glycerol, erythritol, rhamnose, arabitol, sorbitol, galac­ myces cerevisiae; Bakers Yeast type II; and Sigma Chemi­ titol, mannitol, arabinose, glucose, galactose, lactose, cal Co., Cat. No. 45C-2) in the YPD broth. Incubate and ribose, sucrose, raffinose, and maltose to make a sample the culture as described for the standard media. 100 mg/L standard mix solution. Add methanol to this mix at a 10 g/L concentration. Dilute serially with water E. coli Fermentation Broth Culture—Standard Media to final desired concentrations. This study used the fol­ Dissolve LB broth to a concentration of 25 g/L with lowing concentrations: 90, 80, 70, 60, 50, 40, 30, 20, 10, water, heat to a boil, and autoclave for 15 minutes at 8, 6, 4, 2, 1, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.08, 121 psi. LB broth contains 10 g of tryptone, 5 g of yeast 0.06, and 0.04 mg/L. Methanol concentrations ranged extract, and 10 g of sodium chloride per 25 g. Incubate from 9 to 0.004 g/L after dilution. Prepare two stan­ and sample the culture as described for the yeast stan­ dard solutions at concentrations of 1 g/L. One solution dard media. contained 2,3‑butanediol, erythritol, arabitol, galactitol, arabinose, galactose, and ribose. The other solution con­ RESULTS AND DISCUSSION tained glycerol, rhamnose, sorbitol, mannitol, glucose, Selectivity lactose, sucrose, raffinose, and maltose. Dilute these CarboPac MA1 solutions to 0.8, 0.6, 0.4, and 0.2 g/L. Figure 1 shows the separation of alcohols (2,3-butanediol, ethanol, methanol), glycols (glycerol), SAMPLE PREPARATION alditols (erythritol, arabitol, sorbitol, galactitol, mannitol), Yeast Fermentation Broth Culture—Standard Media and carbohydrates (rhamnose, arabinose, glucose, galac­ In a sterile 500-mL Erlenmeyer flask, dissolve 10 tose, lactose, sucrose, raffinose, maltose) commonly found g of Bacto YPD broth (DIFCO Laboratories, Cat. No. in fermentation broths using a CarboPac MA1 column set 0428‑17‑5) in 200 mL of purified sterile water. Bacto with 480 mM sodium hydroxide eluent flowing at YPD broth contains 2 g of Bacto Yeast Extract, 4 g of 0.4 mL/min. The alcohols, sugar alcohols (alditols), gly­ Bacto Peptone, and 4 g of dextrose (glucose) per 10 g. cols, and carbohydrates are well resolved. Maltose elutes Dissolve 1.0 g of yeast (Saccharomyces cerevisiae; Bak­ last at about 60 minutes. ers Yeast type II; Sigma Chemical Co., Cat. No. 45C‑2) in the YPD broth. Cap the flask with a vented rubber stopper. Incubate the culture in a 37 °C shaking water Column: CarboPac MA1, MA1 guard bath (500–600 rpm) for 24 hours. Remove aliquots at Eluent: 480 mM Sodium hydroxide Flow Rate: 0.4 mL/min designated time points and place on ice. Inj. Volume: 10 µL For this study, samples were taken after the addition Detection: Pulsed amperometry, gold electrode Peaks: 1. 2,3-Butanediol 10. Mannitol of yeast at 0, 0.5, 1, 2, 3, 4, 5, 6, 7, and 24 hour intervals. 2. Ethanol 11. Arabinose The incubation began when yeast was added to the me­ 140 3. Methanol 12. Glucose dia. Aliquots were centrifuged at 14,000 g for 5 4. Glycerol 13. Galactose × 5. Erythritol 14. Lactose 10 minutes and diluted 100-fold in purified water. Di­ 4 6. Rhamnose 15. Ribose luted supernatant (10 µL) was analyzed directly. 7. Arabitol 16. Sucrose nC 8. Sorbitol 17. Raffinose 3 9. Galactitol 18. Maltose 2 7 Yeast Fermentation Broth Culture—Modified Multiple 8 910 1 6 12 13 11 15 Carbohydrate Media 14 16 17 18 In a sterile 500-mL Erlenmeyer flask, dissolve 2 g 0

of Bacto Yeast Extract (DIFCO Laboratories, Cat. No. 0 10 3020 40 50 60 70 0127-15-1), 4 g of Bacto Peptone (DIFCO Laboratories, Minutes 13604 Cat. No. 0118-15-2), and 4 g of carbohydrates (0.4 g each Figure 1. Common carbohydrates, alditols, alcohols, and glycols of glucose, sucrose, maltose, lactose, galactose, sorbitol, found in fermentation broths separated on the CarboPac MA1 column with pulsed amperometry.

82 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths The retention times of common fermentation broth CarboPac PA1 carbohydrates under these conditions are listed in Figure 2 shows the analysis of common fermenta­ Table 1. Generally, alcohols elute first, followed by tion broth alcohols, glycols, alditols, and carbohydrates glycols, alditols, monosaccharides, disaccharides, and using the CarboPac PA1 column. The elution order of the trisaccharides. Ethanol, methanol, and 2,3-butanediol are PA1 is similar to the MA1. Alcohols elute first, followed resolved, but the ethanol response is about 570 times less by glycols, alditols, and mono-, di-, and trisaccharides. than the glucose response, and methanol is about At elevated eluent strengths (e.g., 100–250 mM sodium 3600 times less responsive than glucose by mass. This can hydroxide), many larger carbohydrates separate in under be advantageous when relatively large amounts of ethanol 20 minutes. For example, maltotriose, which elutes after or methanol are produced, such as in the manufacture 60 minutes on the MA1, elutes at 25.8 minutes with a of alcoholic beverages and in the generation of alterna­ 250 mM sodium hydroxide eluent on the PA1. Under tive energy sources. Some large carbohydrates such as these conditions, the early eluting peaks (e.g., 2,3-butane­ maltotriose could not be eluted under these conditions and diol, methanol, and ethanol) coelute. are best analyzed using the CarboPac PA1 column. Table 2 summarizes the retention times for common fermentation analytes using the PA1 column at 16, 50, 100, and 250 mM sodium hydroxide eluent conditions Table 1. Retention Times for Carbohydrates and with 1 mL/min flow rates. These results demonstrate that Alcohols on the CarboPac MA1 adjustment of the eluent strength modifies column selectivity, sometimes changing analyte elution order. For Analyte Retention Time (min) example, at 16 mM sodium hydroxide, sucrose eluted 2,3-Butanediol 6.6 slightly before ribose; yet at 50, 100, or 250 mM, ribose Ethanol 7.3 eluted significantly ahead of sucrose. Therefore, some Methanol 7.7 Glycerol 8.7 separations can be improved by adjusting eluent strength. Erythritol 10.7 Adjusting eluent strengh can increase or decrease the Rhamnose 13.6 peak area response of some analytes. Generally, response Fucose 13.8 increases with increased eluent strength, but can decrease Arabitol 14.7 for some analytes. A decrease in response at elevated elu­ Galactosamine 14.7 ent strengths is probably due to hydroxide ions competing Glucosamine 15.3 Sorbitol 16.1 with analytes for sites on the electrode surface. Trehalose 17.0 Galactitol 17.6 Ribitol 17.8 Column: CarboPac PA1, PA1 guard Mannitol 19.4 Eluent: 16 mM Sodium hydroxide 2-Deoxy-D-Glucose 20.2 Flow Rate: 1.0 mL/min Inj. Volume: 10 µL Mannose 21.6 Detection: Pulsed amperometry, gold electrode Arabinose 21.6 2 3 5 Peaks: 1. 2,3-Butanediol/Ethanol 8. Rhamnose 4 Glucose 24.0 1 6 7 2. Glycerol/Methanol 9. Arabinose Xylose 24.6 100 3. Erythritol 10. Galactose Galactose 27.0 4. Arabitol 11. Glucose 5. Galactitol 12. Sucrose Maltitol 27.7 8 9 6. Sorbitol 13. Lactose Lactose 28.9 11 7. Mannitol 14. Raffinose nC 10 Fructose 29.0 12 Ribose 31.5 Cellobiose 43.2 Sucrose 44.9 13 14 Raffinose 51.8 0 Maltose 59.4 Maltotriose >60.0 0 10 20 30 40 50 60 Minutes 13602

Figure 2. Common carbohydrates, alditols, alcohols, and glycols found in fermentation broths separated on the CarboPac PA1 column. 83 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths Method Detection Limits Table 2 CarboPac PA1 with PA1 Guard Retention Times (Minutes) The method detection limits (MDL) for a 10-µL in­ jection of common fermentation broth constituents using Sodium Hydroxide Concentration at 1.0 mL/min the MA1 column are shown in Table 3. The MDL is de­ Analyte 16 mM 50 mM 100 mM 250 mM fined as the minimum concentration required to produce 2,3-Butanediol 1.5 1.5 1.5 1.5 a peak height signal-to-noise ratio of 3. The MDL can Ethanol 1.6 1.6 1.6 1.6 be further decreased by increasing the injection volume Glycerol 1.7 1.7 1.7 1.7 Methanol 1.7 1.7 1.7 1.7 above the 10-µL injection volume used in this Applica­ Erythritol 1.9 1.8 1.8 1.8 tion Note, and by using smoothing algorithms available Arabitol 2.4 2.3 2.2 2.1 in PeakNet software.11–13 Detection limits generally Galactitol 2.7 2.5 2.5 2.3 Ribitol 2.8 2.6 2.5 2.3 increase with longer retention times because of peak Sorbitol 2.8 2.6 2.5 2.2 broadening. Methanol and ethanol have higher detection Mannitol 3.3 3.0 2.8 2.4 limits because their response factors are lower. Trehalose 3.7 3.3 3.1 2.6 Fucose 5.5 4.0 3.2 2.3 Maltitol 8.9 7.1 5.9 3.9 Linearity 2-Deoxy-D-Glucose 9.3 6.2 4.6 3.0 Glycerol, 2,3-butanediol, erythritol, rhamnose, Rhamnose 9.6 5.4 3.8 2.5 arabitol, sorbitol, galactitol, mannitol, arabinose, glu­ Galactosamine 10.7 6.2 4.4 2.7 cose, galactose, lactose, ribose, sucrose, raffinose, and Arabinose 10.9 6.6 4.6 2.9 Glucosamine 12.7 6.9 4.7 2.8 maltose standards ranging from 0.1 to 1000 mg/L (1 to Galactose 14.3 8.4 5.8 3.3 10,000 ng) were injected (n=2 to 3 per concentration) Glucose 15.5 8.7 5.8 3.3 onto a CarboPac MA1 column. Figure 3 shows that the Mannose 16.9 8.6 5.5 3.1 method was linear for 2,3-butanediol, rhamnose, arabitol, Xylose 17.3 9.2 6.0 3.3 Fructose 20.4 10.3 6.5 3.6 sorbitol, mannitol, arabinose, glucose, galactose, lactose, Sucrose 21.7 15.4 10.9 6.0 ribose, sucrose, raffinose, and maltose over this range Ribose 22.0 11.1 7.0 3.8 (r2=0.998–0.999). Glycerol was linear over the range of Lactose 38.5 18.9 10.7 5.0 Raffinose 47.1 31.0 21.1 9.3 Cellobiose >60 31.8 17.7 6.9 Maltose >60 55.7 27.0 9.5 Table 3. Estimated Detection Limits* Using the Maltotriose >60 >60 43.5 25.8 CarboPac MA1 with Pulsed Amperometry

Analyte ng µg/L** The CarboPac PA1 is similar to the CarboPac PA10. 2,3-Butanediol 1 100 The PA10 is solvent-compatible and has better resolution Ethanol 300 30000 between amino and neutral sugars. In some cases, the Methanol 7000 700000 PA10 has a slightly different selectivity. For example, Glycerol 0.4 40 Erythritol 0.2 20 sucrose and fructose coelute on the PA10 at an eluent Rhamnose 1 100 strength of 16–18 mM sodium hydroxide, but are well Arabitol 0.5 50 resolved on the PA1 column. At low eluent strengths, Sorbitol 0.8 80 sucrose and ribose coelute on the PA1 column, but are Galactitol 0.7 70 resolved on the PA10 column, especially at lower so­ Mannitol 0.7 70 dium hydroxide concentrations (10 mM). Arabinose 1 100 Glucose 0.9 90 Figures 1 and 2 and Tables 1 and 2 show that the Galactose 1 100 CarboPac MA1 and PA1 columns have different selec­ Lactose 2 200 tivities and therefore different strengths for determining Ribose 1 100 the alcohols and carbohydrates in fermentation broths. Sucrose 4 400 Column choice will be dictated by the analytes, their Raffinose 5 500 concentrations, and the desired analysis time. Maltose 9 900 *Lower Limit of Detection is Based on 3 X Baseline Noise **10 µL Injections

84 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths 2000000000 2,3-Butanediol; r2=0.9995 Glycerol: r2= 0.9998* 1800000000 Erythritol: r2= 0.9965 Rhamnose: r2= 0.9987 1600000000 Arabitol: r2= 0.9986 2 1400000000 Sorbitol: r = 0.9978 Galactitol: r2= 0.9971 1200000000 Mannitol: r2= 0.9987 Arabinose: r2= 0.9993 1000000000 Glucose: r2= 0.9991 Area Units Galactose: r2= 0.9991 800000000 Lactose: r2= 0.9998 Ribose: r2 = 0.9992 600000000 Sucrose: r2= 0.9987 Raffinose: r2= 0.9987 400000000 Maltose: r2= 0.9991 * Second degree polynomial regression. 200000000 Glycerol is linear (first degree polynomial) from 0.04 to 200 mg/L (r2=0.998) 0 9000800070006000500040003000200010000 10000 14351 ng Injected

Figure 3. Method linearity using the CarboPac MA1 with pulsed amperometric detection.

0.04–200 mg/L (0.04–2000 ng, r2=0.998); erythritol over Figures 4 and 5 show the stability of peak area and the range of 0.04–100 mg/L (0.4–1000 ng, r2=0.999); and retention time for fermentation broths analyzed over galactitol over the range of 0.07–100 mg/L (0.7–1000 ng, 48 hours. At this concentration, peak area RSDs ranged r2=0.998). For the range of 0.1–1000 mg/L (1–10,000 ng), from 2 to 7%, and retention time RSDs ranged from 0.2 glycerol, erythritol, and galactitol deviated from linear­ to 0.4%. Precision is affected by concentration (i.e., RSD ity (r2=0.987, 0.997, 0.997, respectively). Using a second values increase as the concentrations approach the MDL). order polynomial regression, the r2 for glycerol, erythritol, and galactitol was 1.000. For all analytes, linearity was demonstrated over at least three orders of magnitude, and Table 4. Peak Area and Retention Time for most analytes, over four orders of magnitude. Broad Precision over 48 Hours, RSD (%) linear ranges help reduce the need to dilute the sample and repeat the analysis when components vary greatly in Analyte Peak Area Units Retention Time concentration. 2,3-Butanediol 2.4 0.2 Ethanol 2.7 0.2 Precision and Stability Glycerol 2.0 0.2 Peak area and retention time RSDs were determined Erythritol 3.4 0.2 for replicate injections of common carbohydrates, alditols, Rhamnose 1.8 0.2 Arabitol 3.0 0.2 alcohols, and glycols spiked into yeast fermentation broth. Sorbitol 2.7 0.2 Common fermentation broth carbohydrate and alcohol Galactitol 2.7 0.3 standards were added (10 mg/L) to heat-treated yeast Mannitol 2.7 0.3 Arabinose 3.1 0.3 fermentation broth culture supernatant and then analyzed Glucose 3.3 0.2 over 48 hours (10 µL per injection, 42 injections) on Galactose 3.5 0.3 the MA1 column. Results for precision are presented in Lactose 3.6 0.3 Table 4. Ribose 3.1 0.3 Raffinose 4.8 0.4 Maltose 6.8 0.4

85 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths Determination of Carbohydrates, AlditolS, and galactose. Rhamnose, sorbitol, arabinose, lactose, and Alcohols, and Glycols in Fermentation Broth ribose were either unchanged or decreased slightly over Cultures 24 hours. Glycerol increased for the first hour, and then Yeast (Saccharomyces cerevisiae) Culture leveled off. Sucrose could not be measured, even at Yeast was grown in Bacto YPD broth at 37 ˚C for up 0.15 hours (9 min) of incubation, which was the earli­ to 24 hours. Figure 6 shows the separation of fermenta­ est time point possible due to the time required for full tion broth ingredients in a yeast culture at the beginning yeast dissolution. Sucrose was probably digested by the (Figure 6A) and after 24 hours (Figure 6B) of incuba­ extracellular enzyme invertase, which is present in large tion. At the beginning of the culture, the glucose (dex­ amounts in the dried yeast. Invertase will cleave sucrose trose) component was prominent. Ethanol was found at into its monosaccharides, glucose and fructose. Glucose a relatively high concentration, along with trace levels of was measured at levels higher than expected at the first glycerol, erythritol, rhamnose, trehalose, arabinose, and time point, which supports this hypothesis. cellobiose. During the first 3 hours, glucose levels decreased, and after

3 hours glucose was not detected 2,3-Butanediol (data not shown). Ethanol Methanol Glycerol increased over the same Glycerol 40000000 time period and remained constant Erythritol 35000000 Rhamnose after 3 hours. Ethanol concentration Arabitol remained constant up to 7 hours. 30000000 Sorbitol Galactitol 25000000 Between 7 and 24 hours, ethanol Mannitol 20000000 Arabinose concentration decreased, presumably Peak Area Glucose due to evaporative losses. Erythritol 15000000 Galactose Lactose and rhamnose concentrations did 10000000 Ribose not change, cellobiose concentration 5000000 Sucrose Raffinose decreased by 50%, and trehalose and 0 Maltose arabinose were depleted between 0 10 20 30 40 50 60 Time (Hours) 14352 7 and 24 hours. When the cell culture broth Figure 4. Stability of peak area over 48 hours for fermentation broth analysis. Injections 25 was modified to contain ten differ­ through 28 were standards in water. ent carbohydrates and alditols, at the same total carbohydrate concentration as the standard Bacto YPD broth, it was apparent that yeast prefer to use 2,3-Butanediol Ethanol certain carbohydrates over others, and Methanol that some carbohydrates or alditols 70 Glycerol Erythritol could not be used as a carbon source 60 Rhamnose during the 24-hour incubation period. Arabitol 50 Sorbitol Figure 7 shows the concentration of Galactitol 40 Mannitol broth components over 24 hours. Glu­ Arabinose cose and raffinose were metabolized 30 Glucose Galactose within the first hour. After one hour, Retention Time (min) 20 Lactose the yeast began to consume maltose Ribose 10 Sucrose Raffinose 0 Maltose 0 10 20 30 40 50 60 Time (Hours) 14353

Figure 5. Stability of retention time over 48 hours for fermentation broth analysis. Injections 25 through 28 were standards in water.

86 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths E. coli Culture Column: CarboPac MA1, MA1 guard E. coli was grown on LB broth for 24 hours at 37 ˚C. Eluent: 480 mM Sodium hydroxide Flow Rate: 0.4 mL/min Figure 8 shows that only trace levels of carbohydrates Inj. Volume: 10 µL were found in this media. Erythritol, arabitol, arabinose, Detection: Pulsed amperometry, gold electrode Sample: 100-fold dilution, supernatant lactose, and maltose were identified by retention time in Peaks: 1. Unknown 9. Arabitol the starting media. No glucose was measured. After 2. Unknown 10. Trehalose 24 hours, trace levels of 2,3-butanediol, erythritol, manni­ 3. Ethanol 11. Unknown 30 A 13 4. Glycerol 12. Arabinose tol, glucose, and galactose were measured. Many uniden­ 4 5. Unknown 13. Glucose tified peaks were consumed during this incubation period. 3 6. Erythritol 14. Unknown 7. Unknown 15. Cellobiose nC 10 8. Rhamnose 16. Unknown CONCLUSION

2 These results show that both yeast and bacterial 1 9 15 16 6 7 8 12 11 fermentation broths can be analyzed for carbohydrate 0 composition using high-performance anion-exchange chromatography and pulsed amperometry. Two columns

30 (CarboPac MA1 and CarboPac PA1) are available for B 4 the analysis of fermentation broth carbohydrates, alco­ hols, alditols, and glycols. The MA1 provides excellent separation of early eluting compounds such as alcohols, nC glycols, alditols, and monosaccharides. The run times 2 are long for more complex carbohydrates such as trisac­ 1 6 3 5 7 8 14 charides. Separations using the PA1 are faster and can 11 15 16 0 effectively separate di- and trisaccharides. Complex mixtures of carbohydrates and alditols can be monitored 0 10 20 30 40 50 60 simultaneously, providing the analyst with information 13611 Minutes that is needed to optimize the fermentation.

Figure 6. Saccharomyces cerevisiae fermentation broth culture us- ing the CarboPac MA1 column, at (A) 0 hours of incubation and (B) 24 hours of incubation.

3.0

2.5 Rhamnose 2.0 Sorbitol Arabinose Glucose 1.5 Galactose Lactose

Concentration (mg/mL) 1.0 Ribose Raffinose Maltose 0.5 Glycerol

0 0.15 0.5 1 2345678924 14354 Incubation Time (Hours)

Figure 7. Saccharomyces cerevisiae culture grown on fermentation broth consisting of multiple carbohydrates and alditols, analyzed on a CarboPac MA1 with integrated amperometry.

87 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths 5. Rank, M.; Gram, J.; Danielsson, B. Anal. Chim. Column: CarboPac MA1, MA1 guard Eluent: 480 mM Sodium hydroxide Acta. 1993, 281, 521–526. Flow Rate: 0.4 mL/min 6. Buttler, T.; Gordon, L.; Marko-Varga, G. Anal. Inj. Volume: 10 µL Chim. Acta. 1993, 279, 27–37. Detection: Pulsed amperometry, gold electrode Sample: 10-fold dilution, supernatant 7. van de Merbel, N.C.; Lingeman, H.; Brinkman, Peaks: 1. Unknown 13. Galactitol U.A. Th.; Kolhorn, A.; de Rijke, L.C. Anal. Chim. 2. Unknown 14. Unknown 3. 2,3-Butanediol 15. Arabinose Acta 1993, 279, 39–50. 4. Ethanol 16. Unknown 8. van de Merbel, N.C.; Kool, I.M.; Lingeman, H.; 5. Glycerol 17. Lactose Brinkman, U.A. Th.; Kolhorn, A.; de Rijke, L.C. 6. Unknown 18. Unknown 7. Erythritol 19. Unknown Chromatographia 1992, 33 (11/12), 525–532. 8. Unknown 20. Unknown 50 A 5 9. Marko-Varga, G.; Dominguez, E.; Hahn-Hagerdal, 9. Rhamnose 21. Unknown 10. Arabitol 22. Maltose B.; Gorton, L.; Irth, H.; De Jong, G.J.; Frei, R.W.; 12 11. Unknown 23. Unknown Brinkman, U.A. Th. J. Chromatogr. 1990, 523, 14 16 12. Unknown 173–188. nC 10. Dionex Corporation, Application Note 123, “Deter­ 20 9 mination of organic and inorganic anions in fermen­ 7 14 3 6 17 10 19 tation broths”. 8 15 18 21 22 23 0 11. Dionex Corporation, Technical Note 40, “Glyco­ protein monosaccharide analysis using high-perfor­ 50 B 1 mance anion-exchange chromatography with pulsed 2 amperometric detection (HPAE-PAD)”. 12. Schibler, J.A. Am. Lab. 1997, 52–54.

4 13. Dionex Corporation, Technical Note 43, nC 5 20 “Using smoothing algorithms to reduce baseline 3 11 noise in chromatography”. 7 9 14 13 6 8 15 21 23 0 LIST OF SUPPLIERS Aldrich Chemical Company, Inc., 1001 West Saint Paul 0 10 3020 40 50 60 70 Minutes 13615 Avenue, P.O. Box 355, Milwaukee, Wisconsin, 53233, USA. Tel.: 1-800-558-9160.

Figure 8. E. coli fermentation broth culture using the CarboPac J.T. Baker Incorporated, 222 Red School Lane, Phillips­ MA1 column, at (A) 0 hours of incubation and (B) 24 hours of burg, New Jersey, 08865, USA. incubation. Tel.: 1-800-582-2537. Eastman Chemical Company, 1001 Lee Road, Roches­ ter, New York, 14652-3512, USA. Tel.: 1-800-225- REFERENCES 5352. 1. Robinett, R.S.R.; Herber, W.K. J. Chromatogr. A. EM Science, P.O. Box 70, 480 Democrat Road, Gibb­ 1994, 671, 315–322. stown, New Jersey, 08027, USA. 2. Herber, W.K.; Robinett, R.S.R. J. Chromatogr. A. Tel.: 1-800-222-0342. 1994, 676, 287–295. Fisher Scientific, 711 Forbes Avenue, Pittsburgh, Penn­ 3. Marko-Varga, G.; Buttler, T.; Gorton, L.; Olsson, sylvania, 15219-4785, USA. Tel.: 1-800-766-7000. L.; Durand, G.; Barcelo, D. J. Chromatogr. A. 1994, Pfanstiehl Laboratories, Inc., 1219 Glen Rock Avenue, 665, 317–332. Waukegan, Illinois, 60085-0439, USA. 4. Schugerl, K.; Brandes, L.; Wu, X.; Bode, J.; Ree, Tel.: 1-800-383-0126. J.I.; Brandt, J.; Hitzmann, B. Anal. Chim. Acta. Sigma Chemical Company, P.O. Box 14508, St. Louis, 1993, 279, 3–16. Missouri, 63178, USA. Tel.: 1-800-325-3010.

88 The Determination of Carbohydrates, Alcohols, and Glycols in Fermentation Broths Application Brief 136

Determination of Inorganic Counterions in Pharmaceutical Drugs Using Capillary IC

INTRODUCTION many benefits to IC users. Thermo Scientific Dionex One of the most important applications of ion Capillary Reagent-Free™ IC (RFIC™) systems deliver fast chromatography (IC) is to determine counterions in active results by reducing eluent preparation, system startup, and pharmaceutical ingredients (API) and drug products in equilibration times. Perhaps most importantly, the system the pharmaceutical industry.1,2 Approximately 50% of all can be left on (i.e., running), always ready for analysis drugs on the market are developed in salt forms.3,4 Certain because of its low consumption of eluent (15 mL a day). suboptimal physicochemical and biopharmaceutical Having the system always on and ready significantly properties of APIs can be overcome by pairing a basic streamlines the IC workflow. An always on system or acidic drug molecule with a counterion to create a salt maintains stability and requires less frequent calibrations. version of the drug with high solubility, stable crystalline The amount of waste generated is significantly decreased form, and good bioavailability. Ion chromatography with and the Thermo Scientific Dionex EluGen cartridge suppressed conductivity detection plays an important role producing the eluent lasts 18 months under continuous in the salt selection process to establish correct molecular operation mode, which translates into reduced overall mass of the entity in early stages of drug development. cost of ownership. Ion chromatography can also be used in quality control to Figure 1 shows the analysis of chloride in a drug verify identity, strength, and purity of ionic APIs. used to treat type 2 diabetes using the Thermo Scientific This study describes the determination of inorganic Dionex IonPac™ AS19 capillary column designed for anions and cations in two different drugs using the diverse sample matrices. This column is ideally suited capillary Thermo Scientific Dionex ICS-5000 system. for use with the RFIC system. The analysis time for this Scaling down from standard bore to capillary scale brings counteranion is less than 5 min.

89 Determination of Inorganic Counterions in Pharmaceutical Drugs Using Capillary IC Column: Dionex IonPac AS19, Capillary, 0.4 × 250 mm Column: Dionex IonPac CS12A, Capillary, 0.4 × 250 mm Eluent Source: Dionex EGC-KOH (Capillary) Eluent Source: Dionex EGC-MSA (Capillary) Eluent: 40 mM KOH Gradient: 20 mM MSA Flow Rate: 10 µL/min Flow Rate: 10 µL/min Inj. Volume: 0.4 µL Inj. Volume: 0.4 µL Column Temp.: 30 °C Column Temp.: 30 °C ™ Detection: Suppressed conductivity, Thermo Scientific Dionex ACES 300 Detection: Suppressed conductivity, Thermo Scientific Dionex Anion Capillary Electrolytic Suppressor CCES™ 300 Cation Capillary Electrolytic Suppressor Sample: 100 mg strength tablet Sample: 10 mg strength tablet Sample Prep.: Thermo Scientific Dionex OnGuard RP cartridge, filtered, 1:10 dilution Sample Prep.: Dionex OnGuard RP cartridge, filtered, 1:10 dilution mg/L mg/tablet mg/L µg/tablet Peak: 1. Chloride 16.6 0.453 1.8 Peaks: 1. Sodium 0.147 1.13 2. Magnesium 0.705 5.47 18 1 3. Calcium 3.59 27.8

1

µS µS 3

2

-1 -0.2 0 1 263 4 5 7 0 5 10 15 Minutes Minutes 28888 28889

Figure 1. Determination of a counteranion in a drug used to treat Figure 2. Determination of a countercation in a drug used to type 2 diabetes. control cholesterol.

Analysis of the countercation calcium in a drug used REFERENCES to control cholesterol is illustrated in Figure 2. Using the 1. Dionex Corporation. Quantification of Anions in Dionex IonPac CS12A column, calcium is well separated Pharmaceuticals. Application Note 116, LPN 0924, from sodium and magnesium present in the excipients. 2007, Sunnyvale, CA. 2. Dionex Corporation. Determination of EQUIPMENT AND CONDITIONS Sulfate Counter Ion and Anionic Impurities in The Dionex Capillary ICS-5000 system, Thermo Aminoglycoside Drug Substances by IC with Scientific Dionex AS-AP Autosampler, and Thermo Suppressed Conductivity Detection. Application Scientific Dionex Chromeleon™ software are used in Note 190, LPN 1946, 2007, Sunnyvale, CA. this experiment. All experimental parameters are listed 3. Wermuth, C.G.; Stahl, P.H. Handbook of in the figures above. Pharmaceutical Salts: Properties, Selection and Use; Wiley-VCH: Weinheim, Germany, 2002, pp 1–7. SAMPLE PREPARATION 4. Berge, S.M.; Bighley, L.M.; Monkhouse, D.C. Extract the counterion analyte by dissolving the Pharmaceutical Salts. J. Pharm. Sci. 1977, tablet in water after 50 °C. Treat the sample using the 66 (1), 1–19. Dionex OnGuard™ RP cartridge, then filter through a 0.4 µm syringe filter, and dilute the sample solution 10-fold prior to analysis.

90 Determination of Inorganic Counterions in Pharmaceutical Drugs Using Capillary IC Application Note 233

Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column

INTRODUCTION of glucosamine and uronic acid. Acid hydrolysis of In early 2008, unusual reactions to heparin, including heparin samples releases glucosamine, which is easily hypotension, swelling of the larynx, and in some cases determined by electrochemical detection. In comparison, death, prompted a recall of the product.1,2 The lots of chondroitin sulfates are composed of galactosamine and heparin that led to these anaphylactic reactions were uronic acid. In these compounds, acid hydrolysis releases tested using existing reported methods and no additional galactosamine, which can also easily be determined by components were found. After extensive investigation, electrochemical detection. The USP method measures it was determined that the heparin in question had been the ratio of galactosamine/glucosamine as an indication contaminated with oversulfated chondroitin sulfate.3 of the heparin purity and to identify heparin samples that At the time of the recall, existing heparin assay may be contaminated or adulterated with chondroitin methods could not detect chondroitin sulfates in heparin. sulfate compounds. For this reason, the U.S. Pharmacopeia (USP) has In this AN, the organic impurities in heparin proposed a revision to the heparin sodium monograph that are determined by the HPAE-PAD method using the includes chromatographic methods for the identification CarboPac® PA20 column following the USP monograph of heparin and the determination of organic impurities in method. This method was repeated using manually heparin.4 This USP monograph is scheduled to become prepared eluents and an electrolytically generated eluent, official on August 15, 2009. The heparin chromatographic with both eluent preparation options providing data identity portion of the monograph, which determines that exceeds the system suitability requirements in the oversulfated chondroitin and dermatan sulfate in heparin, monograph. In addition, this document includes deliberate will not be discussed in this Application Note (AN), but is variation of several chromatographic parameters to available elsewhere.5 evaluate method ruggedness. The HPAE-PAD method The organic impurities section of the heparin provides sensitive determination of galactosamine in monograph relies on hydrolyzing the polysaccharide and acid- hydrolyzed heparin samples, enabling the determining the relative amounts of galactosamine and identification of heparin that has been contaminated glucosamine in the sample digests. Heparin is composed with chondroitin sulfates.

91 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column EQUIPMENT SAMPLES Dionex ICS-3000 Reagent-Free™ IC system with eluent Chondroitin sulfate B, sodium salt (β-heparin, dermatan generation (RFIC-EG™) system consisting of: sulfate, sodium salt) (Sigma-Aldrich P/N C3788) SP Single Pump or DP Dual Pump module Sample A: Heparin, sodium salt, grade 1-A (Gradient pump required if manual eluent is used) (Sigma-Aldrich P/N H3393) EG Eluent Generator module Sample B: Heparin, sodium salt (Sigma-Aldrich DC Detector/Chromatography module (single or dual P/N H4784) temperature zone configuration) CONDITIONS AS Autosampler Columns: AminoTrap™ column EluGen® EGC II KOH cartridge (Dionex P/N 058900 ) 3 × 30 mm (P/N 060146) Continuously-Regenerated Anion Trap Column, CarboPac PA20 Analytical, CR-ATC (Dionex P/N 060477) 3 × 150 mm (P/N 060142) ICS-3000 ED Electrochemical detector -or- (Dionex P/N 079831) AminoTrap column, Electrochemical cell (Dionex P/N 061757) 3 × 30 mm (P/N 060146) Disposable gold electrode, carbohydrate certified CarboPac PA20 Guard, (Dionex P/N 060139) 3 × 30 mm (P/N 060144) Reference electrode (Dionex P/N 061879) CarboPac PA20 Analytical, 3 × 150 mm 10 µL PEEK™ Sample injection loop (Dionex P/N 042949) (P/N 060142) (manual eluent only) EG Vacuum Degas Conversion Kit (Dionex P/N 063353) Eluent: 14 mM KOH from -10–0 min, ® Chromeleon 6.8 Chromatography Data System 14 mM KOH from 0–10 min, 0.3 mL polypropylene injection vials with caps 100 mM KOH from 10–20 min (Dionex P/N 055428) Eluent Source: EGC II KOH with CR-ATC ® Nalgene 1000 mL 0.2 µm nylon filter units -or- (VWR P/N 28198-514) 200 mM KOH, manually prepared 7 mL polypropylene screw cap tubes Flow Rate: 0.5 mL/min (Sarstedt P/N 60.550) Temperature: 30 °C 500 mL PMP volumetric flasks, Class A Inj. Volume: 10 µL (Vitlab P/N 67504) Detection: Pulsed amperometric, Dry block heater (VWR P/N 13259-005) disposable gold working electrode REAGENTS AND STANDARDS Background: 5–25 nC (using the carbohydrate waveform) Deionized water (DI), Type I reagent grade, 18 MΩ-cm resistivity or better Noise: 20–50 pC Hydrochloric acid, Ultrex II, (JT Baker P/N 6900-05) System Potassium hydroxide, 45% (w/w) (Fisher P/N SP236-500) Backpressure: ~2625 psi (using the AminoTrap 3 × 30 mm and Sodium hydroxide, 50% (w/w) (Fisher P/N SS254-500) CarboPac PA20 3 × 150 mm columns) Glucosamine hydrochloride (Sigma-Aldrich P/N G-4875) -or- Galactosamine hydrochloride (Pfanstiehl Laboratories ~3010 psi (using the AminoTrap P/N RGG-104) 3 × 30 mm, CarboPac PA20 3 × 30 mm guard, and CarboPac PA20 3 × 150 mm analytical columns as described by the USP)

92 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column Carbohydrate 4-Potential Waveform for the ED Time(s) Potential(V) Gain Region* Ramp* Integration Standard Stock Solutions 0.00 +0.1 Off On Off Glucosamine 0.20 +0.1 On On On Prepare a 1.6 mg/mL stock solution of glucosamine 0.40 +0.1 Off On Off hydrochloride by dissolving 0.1600 g of glucosamine 0.41 –2.0 Off On Off hydrochloride in 100 mL of 5 N hydrochloric acid. 0.42 –2.0 Off On Off This stock will be used to prepare the standard solution 0.43 +0.6 Off On Off for digestions. 0.44 –0.1 Off On Off 0.50 –0.1 Off On Off Galactosamine *Settings required in the ICS-3000, but not used in older Prepare a 16 mg/mL solution of galactosamine Dionex systems. hydrochloride by dissolving 0.0320 g of galactosamine Reference electrode in Ag mode (Ag/AgCl reference). hydrochloride in 2.00 mL of DI water. Further dilute See Technical Note 21 for more information.6 this concentrate by adding 100 µL to 99.9 mL of 5 N hydrochloric acid to prepare a 16 µg/mL stock solution PREPARATION OF SOLUTIONS AND REAGENTS of galactosamine. Eluent Solutions Glucosamine and galactosamine stock solutions were Generate the potassium hydroxide (KOH) eluent stored at 4 °C. online by pumping high-quality degassed, DI water through the EGC II KOH cartridge. The Chromeleon Standard Solution software will track the amount of KOH used and calculate Prepare the standard solution by transferring 2.5 mL the remaining lifetime. of glucosamine stock solution into a 7 mL screw cap The method can be executed with manually prepared vial containing 2.5 mL of galactosamine stock solution. eluents. Prepare 1 L of 200 mM KOH from 45% w/w This solution contains 8 µg/mL of galactosamine and KOH concentrate by adding 17 mL of 45% KOH to 800 µg/mL of glucosamine (1% w/w galactosamine with 983 g of degassed, DI water. If desired, NaOH can be respect to glucosamine). Freshly prepare the standard used in place of KOH. To prepare 1 L of 200 mM NaOH, solution before each digestion. add 10.4 mL of 50% NaOH to 989.6 mL of degassed, DI water. DIGESTION OF SAMPLES Proportion the 200 mM hydroxide solution with DI Prepare samples for digestion by adding 12 mg of water to produce either a 14 mM NaOH or KOH eluent heparin to a 7 mL screw cap vial. Add 5 mL of 5 N HCl to for the sample elution or a 100 mM NaOH or KOH the vial and vortex the solution to mix. Heat the samples eluent for column cleaning. See Tech Note 71 for detailed and the standard solution at 100 °C for 6 h to hydrolyze information on manual eluent preparation.7 the samples into glucosamine and galactosamine. After digestion of the samples, allow the samples to cool to 5 N Hydrochloric Acid for Sample Digestion ambient temperature, quantitatively transfer the contents Dilute 102 g (88 mL) of 33% hydrochloric acid to a of the vial to a 500 mL PMP volumetric flask, and fill to total of 211 g with DI water. the mark on the flask with DI water.

93 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column PRECAUTIONS and Experimental CONSIDERATIONS Guard Column Considerations Labware When implementing the method using an EG eluent, Glass volumetric flasks should not be used for the AminoTrap column should be used in place of the dilution of samples and standards after digestion. CarboPac PA20 guard column. If both the CarboPac PA20 Peak heights may be reduced if glass is used. For this Guard and Analytical column are installed, excessive application, Class A PMP flasks were used, although backpressure can occur and the EG cartridge may be polypropylene would be acceptable. Similarly, damaged. To prevent such damage, the Chromeleon polypropylene, rather than glass, digestion vials and software will automatically turn off the pump at a pressure injection vials should be used. of 3000 psi. When using PMP or polypropylene labware, it is When implementing the method using manually important to remove bubbles from the surface of the prepared eluents, the two guard columns and the plastic labware. This can be accomplished by gently CarboPac PA20 column can be used as described in the swirling the solution in the volumetric flask while it is USP monograph. However, only one guard is necessary. approximately one-half full. The final dilution should be If no amino acids are expected in the samples, the made by gently adding water down the side of the flask. CarboPac guard column should be used without the Bubbles on the walls of the flask can lead to dilution AminoTrap column. errors. Similarly, bubbles in injection vials should be When using the AminoTrap column, particular tapped out before the samples are loaded in the AS to care should be taken to avoid flowing deionized water ensure consistent injection volumes. through the trap column. If the column is damaged by water, fronting of the glucosamine and galactosamine Use of Sodium Hydroxide for Manual Eluent Preparation peaks may be observed. This will reduce the resolution Sodium hydroxide can be substituted for potassium between glucosamine and galactosamine and decrease hydroxide when manually preparing eluents. Glucosamine the measured column efficiency. If either peak fronting or peak asymmetry and resolution between galactosamine a sudden decrease in resolution is observed, replace the and glucosamine pass the USP requirements. AminoTrap column.

Equilibration of the Column and Retention Time Precision RESULTS AND DISCUSSION To optimize retention time precision, each sequence Separation should start with 3–5 injections of a 50 mM HCl blank. Figure 1 shows the separation of hydrolyzed When using EG, equilibration of the system with three glucosamine/galactosamine standard solution when blank acid injections led to retention time precision RSDs using the AminoTrap and CarboPac PA20 analytical ranging from <0.001 to 0.58. If greater retention time columns with eluent generation. The galactosamine precision is needed, additional blank injections can be (GalN) peak is well resolved (USP resolution = 3.2) from performed to stabilize the system or the equilibration the glucosamine (GlcN) peak and clearly identified at time prior to sample injection can be increased. When a concentration of 1% of the GlcN concentration. The using manually prepared eluents, this effect is magnified average retention times for GlcN and GalN are 6.51 demonstrating RSDs for glucosamine ranging between and 5.51 min, respectively. Equivalent chromatography <0.01 and 3.1. is obtained if manual eluents are used with the three columns specified in the USP monograph. However, due to the addition of the CarboPac PA20 guard column, the retention times for GlcN and GalN increase to 7.07 and 5.97 min respectively.

94 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column Column: AminoTrap, 3 × 30 mm, CarboPac PA20 3 × 150 mm Column: AminoTrap, 3 × 30 mm, CarboPac PA20 3 × 150 mm Eluent: 14 mM KOH from -10–0 min, 14 mM KOH from 0–10 min, Eluent: 14 mM KOH from -10–0 min, 14 mM KOH from 0–10 min, 100 mM KOH from 10–20 min (using EG) 100 mM KOH from 10–20 min (using EG) Temperature: 30 °C Temperature: 30 °C Flow Rate: 0.5 mL/min Flow Rate: 0.5 mL/min Inj. Volume: 10 µL Inj. Volume: 10 µL Detection: PAD, Au (Disposable) Detection: PAD, Au (disposable) Sample Prep.: Acid hydrolysis Sample Prep.: Acid hydrolysis Samples: Galactosamine/glucosamine standard solution Samples: Heparin sample A spiked with 1% dermatan sulfate (day 1) Peaks: 1. Galactosamine 80 ng/mL Peaks: 1. Galactosamine 1.29 % 2. Glucosamine 8000 2. Glucosamine 98.7 190 150 2 2

nC nC

1 1

10 10 0 2.5 5.0 7.5 10 0 2.5 5.0 7.510 Minutes Minutes 26128 26129

Figure 1. Separation of the standard solution on the Figure 2. Separation of acid-hydrolyzed heparin spiked with CarboPac PA20. 1% dermatan sulfate on the CarboPac PA20.

Sample Analysis The response ratio of GalN/GlcN determined for the Calculations: The percentage of GalN in the heparin standard solutions ranged from 1.03 to 1.19 during four digests is calculated by comparison against the standard weeks of sample analysis. solution, which contains 1% (w/w) of GalN/GlcN in The response factor was used to calculate the 5 N HCl. For each set of digestions a 5 mL aliquot of percentage of galactosamine in the heparin digests standard solution was also digested and the relative according to the formula below: response of GalN/GlcN was calculated by the % GalN = [GalN /GalN ] / following formula: U R [(GalNU/GalNR) + GlcNU] × 100 Response ratio: Where: GalN = the galactosamine peak area from the (GalNR) = (GalNB)/(GalNW) * (GlcNW)/(GlcNB) U Where: hydrolyzed heparin sample GalN = the response ratio GalNB = the galactosamine peak area from the R

hydrolyzed standard solution GlcNU = the glucosamine peak area from the hydrolyzed heparin sample GalNW = the weight of galactosamine in the standard solution Figure 2 shows the separation of heparin sample A GlcN = the weight of glucosamine in the standard W spiked with 1% (w/w) of dermatan sulfate (chondroitin solution sulfate B). The GalN peak is well resolved from GlcN GlcN = the glucosamine peak area from the B (USP resolution = 3.2) and 1.29% GalN was determined hydrolyzed standard solution in the sample. In unspiked samples, 0.04% galactosamine was determined in the hydrolysate.

95 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column Table 1. Comparison of Triplicate Heparin Analysis Results to USP Criteria When Using an EG Eluent* % GalN RSD for Resolution Efficiency Asymmetry Standard Solution Day Sample (USP limit % GalN (USP limit (USP limit (USP limit Response Factor <1%) Determined NLT 2) NLT 2000) 0.8–2.0) Standard solution N/A N/A 3.2 5292 1.1 Heparin, sample A 0.04 3.3 3.2 4955 1.2 1 1.16 1% Dermatan-spiked 1.29 0.17 3.2 5053 1.2 heparin, sample A Standard solution N/A N/A 3.2 5326 1.1 Heparin, sample A 0.04 9.3 3.3 5224 1.1 2 1.19 1% Dermatan-spiked 1.28 0.07 3.2 5330 1.1 heparin sample A Standard solution N/A N/A 3.2 5320 1.1 Heparin, sample A 0.04 9.0 3.3 4441 1.4 3 1.16 1% Dermatan-spiked 1.40 0.79 3.1 4602 1.4 heparin, sample A * AminoTrap and CarboPac PA20 analytical columns used.

Precision and Reproducibility when Using PA20 guard column was installed to match the column Eluent Generation set specified by the proposed method. Table 2 shows the Table 1 displays the USP criteria and the analysis results for the standard solution, heparin sample experimental results for three days of triplicate testing A, dermatan-spiked heparin sample A, and heparin sample of the standard solution, heparin sample A, and the B using both manually prepared KOH and NaOH. In dermatan-spiked heparin. As shown in Table 1, all USP both cases, the USP criteria are met and the determined criteria are met. The between-day sample analysis had an percentages of GalN are consistent. Comparison of these RSD of 0.6, although the intraday precision RSDs ranged percentages with those found while using EG (Table 1) from 3.2 to to 9.3. This precision is excellent considering show that both EG and manual eluent preparation are the low concentrations of galactosamine present in the suitable for the method described in the USP monograph. digested heparin. The value of 0.04% GalN is at the limit of detection and is therefore an extreme measure Column Reproducibility of reproducibility. Spiked heparin showed an average of For comparison, a second column was tested using 1.3% galactosamine with a between-day precision RSD manually prepared KOH eluent. The efficiency of the of 4.3. The differences observed in the spiked samples are column was slightly lower than the original column likely due to slight variations in spiked amounts. used, but results still greatly exceed the USP criteria and analysis of samples led to equivalent results. Table 2 Method Ruggedness shows the results of sample analysis using the same batch Manually Prepared Eluents of manually prepared KOH on two different columns. Manual potassium hydroxide and manual sodium hydroxide eluents were prepared to compare analysis Guard Column Use (Manual Eluents) results against sequences generated using an EG eluent. As a further test, the AminoTrap column was Manually prepared 200 mM potassium or sodium removed. When sample digests were analyzed with hydroxide was prepared and proportioned at 7% and this column set and manual KOH eluent, the % GalN 50% with DI water to deliver 14 mM hydroxide and 100 determined was 0.11%, 0.57%, and 1.40% in heparin mM hydroxide, respectively. Additionally, the CarboPac sample A, heparin sample B, and dermatan-spiked heparin

96 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column Table 2: Comparison of Triplicate Heparin Analysis Results to USP Criteria When Using Manually Prepared KOH or NaOH Eluents* % GalN Resolution Efficiency Asymmetry Instrumental Standard Solution Sample (USP limit (USP limit (USP limit (USP limit Conditions Response Factor <1%) NLT 2) NLT 2000) 0.8–2.0) Standard solution N/A 3.3 5736 1.1 Manually prepared KOH proportioned to 14 mM Heparin, sample A 0.03 3.6 6136 1.2 CarboPac PA20 1% Dermatan-spiked heparin, 1.15 1.40 3.6 6326 1.1 sample A Column 1 Heparin, sample B 0.55 3.6 6359 1.1 Standard solution N/A 3.1 5120 1.2 Manually prepared KOH Heparin, sample A 0.03 3.3 5304 1.2 proportioned to 14 mM 1% Dermatan-spiked heparin, 1.12 1.30 3.2 5353 1.2 Column 2 sample A Heparin, sample B 0.52 3.2 5382 1.2 Standard solution N/A 3.2 5121 1.2 Manually prepared NaOH Heparin, sample A 0.04 3.3 5209 1.2 proportioned to 14 mM 1% Dermatan-spiked heparin, 1.12 1.27 3.2 5206 1.2 Column 2 sample A Heparin, sample B 0.53 3.2 5217 1.2 *AminoTrap, CarboPac PA20 guard, and CarboPac PA20 analytical columns used. sample A, respectively. All USP criteria are exceeded with List of Suppliers the resolution, column efficiency, and peak asymmetry VWR,1310 Goshen Parkway, West Chester, PA 19380, being equivalent to results while using the AminoTrap USA. Tel: 800-932-5000. column. A slight increase in the sensitivity at low http://www.vwr.com concentrations of GalN is observed, but otherwise results Fisher Scientific, One Liberty Lane, Hampton, NH, are equivalent to using all three columns. If samples are 03842, USA. Tel: 800-766-7000 not expected to contain amino acid contaminants, only the http://www.fishersci.com CarboPac PA20 guard is necessary. Sigma-Aldrich, P.O. Box 14508, St. Louis, MO 63178, USA. Tel: 800-325-3010. CONCLUSION http://www.sigma-aldrich.com In this AN, the organic impurities in two research grade heparin samples were determined by the HPAE- PAD system using the CarboPac PA20 column, following the organic impurities method in the proposed revision of the heparin sodium USP monograph. The method ruggedness was shown by comparing results when using EG or manual hydroxide eluents, evaluating the guard columns for both EG and manual eluent system configurations, and by performing sample analysis on two different CarboPac PA20 columns. The HPAE-PAD system allows reliable determination of galactosamine in acid-hydrolyzed heparin samples, thereby providing a method to easily identify heparin that has been contaminated with chondroitin sulfates.

97 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column REFERENCES 4. Heparin Sodium, Pharmocopeial Forum 2009, 35, 1. Contaminant Detected in Heparin Material of 1–10. Specified Origin in the USA and in Germany; Serious 5. Dionex Corporation. Determination of Oversulfated Adverse Events Reported; Recall Measures Initiated. Chondroitin Sulfate, Dermatan Sulfate, and Heparin World Health Organization Alert No. 118, March 7, Sodium Using Anion-Exchange Chromatography with 2008. http://www.who.int/medicines/publications/ UV Detection (IC-UV), Application Note 235, LPN drugalerts/Alert_118_Heparin.pdf. Last accessed 2306. Sunnyvale, CA, 2009. 05/04/09. 6. Dionex Corporation. Optimal Settings for Pulsed 2. Recall of Heparin Sodium Injection and Heparin Amperometric Detection of Carbohydrates Using the Lock Flush Solution (Baxter) FDA Public Health Dionex ED40 Electrochemical Detector, Technical Update, February 2, 2008. http://www.fda.gov/cder/ Note 21, LPN 034889-03. Sunnyvale, CA, 1998. drug/infopage/heparin/public_health_update.htm. 7. Dionex Corporation. Eluent Preparation for High- Last accessed 05/18/09. Performance Anion-Exchange Chromatography with 3. Guerrini, M.; Beccati, D.; Shriver, Z.; Naggi, A.; Pulsed Amperometric Detection, Technical Note 71, Viswanathan, K.; Bisio, A.; Capila, I.; Lansing, LPN 1932-01. Sunnyvale, CA, 2007. J.C.; Guglieri, S.; Fraser, B.; Al-Hakim, A.; Gunay, N.S.; Zhang, Z.; Robinson, L.; Buhse, L.; Nasr, M.; Woodcock, J.; Langer, R.; Venkataraman, G.; Linhardt, R.J.; Casu, B.; Torri, G.; Sasisekharan, R.; Oversulfated Chondroitin Sulfate is a Contaminant in Heparin Associated with Adverse Clinical Events, Nature Biotechnology 2008, 26, 669–675.

98 Determination of Galactosamine Containing Organic Impurities in Heparin by HPAE-PAD Using the CarboPac PA20 Column Application Note 235

Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection

INTRODUCTION sulfate.8 In addition, the USP added a high-performance Heparin, a complex sulfated glycosaminoglycan, is a anion-exchange with pulsed amperometric detection well-known anticoagulant used to stop or prevent blood method (HPAE-PAD) to identify the percent of organic from clotting in certain types of surgeries and dialysis impurities (discussed in Application Note 233).9 The treatments.1–3 In January 2008, the FDA was notified Stage 2 monograph is scheduled to be implemented in that patients using the heparin-active pharmaceutical August 2009. ingredient experienced higher incidences of severe This Application Note (AN) describes the separation allergic reactions, including difficulty in breathing, of dermatan sulfate and OSCS from heparin using nausea, vomiting, hypotension, and even death. An FDA anion-exchange chromatography (AE) with UV detection investigation revealed that the heparin was adulterated according to the USP Stage 2 revision. Dermatan sulfate, with a highly sulfated version of chondroitin sulfate (a OSCS, and heparin sodium are separated on a galactosamine glycosaminoglycan), subsequently referred 2 mm IonPac® AS11 anion-exchange column (USP L61 to as oversulfated chondroitin sulfate (OSCS).4,5 type column) using a gradient separation, with sodium The assay for heparin sodium in the U.S. phosphate monobasic and sodium perchlorate mobile Pharmacopeia (USP) monograph measured heparin phase solutions at 0.22 mL/min and detected by UV anti-clotting activity.6 This assay could not detect OSCS, absorbance at 202 nm. This method takes advantage of which is believed to be synthetic. Therefore, the USP the hydrophilic character and low capacity of the published a revision of the monograph (Stage 1 revision, IonPac AS11 column stationary phase to separate 1% USP 32) to address the immediate health issue by (v/v) concentrations of dermatan sulfate and OSCS from identifying OSCS in heparin by capillary electrophoresis a large, single heparin peak. The column diameter and (CE).7 Based on the public comments received from the low flow rate minimizes mobile phase consumption and Stage 1 revision, the USP replaced the CE method with therefore, waste. The method ruggedness and precision an anion-exchange chromatography method are discussed in this AN. This method can detect <1% (Identification B) in the Stage 2 revision. The (v/v) dermatan sulfate and OSCS in adulterated chromatography method improves the resolution of heparin samples. heparin from impurities, such as OSCS and dermatan

99 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection Equipment reagents and standards Dionex ICS-3000 Chromatography system consisting of: Deionized water, Type 1 reagent-grade, SP Single Pump or DP Dual Pump module with a 18.2 MΩ-cm resistivity GM-4 (P/N 049135, 2 mm) gradient mixer column Use ACS reagent grade chemicals for all reagents and TC Thermal Compartment with a 6-port standards, unless otherwise specified. injection valve Heparin Sodium Identification Standard (9.3 mg; VWD Variable Wavelength Detector 3400 with USP P/N 1304038) PEEK™ semi-micro flow cell, 2.5 µL, 7 mm Heparin Sodium System Suitability Standard (>90% (P/N 6074-0300) heparin, <10% oversulfated chondroitin sulfate; AS Autosampler with Sample Tray Temperature 50 mg; USP P/N 1304049) Controlling option, 100 µL sample syringe Chondroitin sulfate B (dermatan sulfate, sodium salt), (P/N 055064), and 1.5 mL sample tray sodium salt from porcine intestinal mucosa, >90% Chromeleon® 6.8 Chromatography Data System lyophilized powder (Sigma-Aldrich P/N C3788)

Bottles, 1 L or 2 L (two each), glass coated, Sodium perchlorate, monohydrate (NaClO4 • H2O, GL45 (Dionex P/N 045900, 045901) for mobile VWR International P/N EM-SX0693) phase solutions Sodium phosphate, monobasic monohydrate

Vial Kit, 1.5 mL glass with caps and septa (P/N 055427) (NaH2PO4 • H2O, VWR International P/N JT3818) or 0.3 mL polypropylene sample vials with caps and Sodium phosphate, monobasic dihydrate slit septa (P/N 055428) (NaH2PO4 • 2H2O, VWR International P/N JT3819) Nalgene® Media-Plus with 90 mm, 0.45 µm nylon filter Phosphoric acid, 85–87% (H3PO4, VWR International (Nalge Nunc International P/N 164-0020) or P/N JT0260) equivalent nylon filter pH 7 (yellow) buffer solution (VWR International Vacuum pump P/N BDH5046) to calibrate the pH meter PEEK Tubing: pH 4 (red) buffer solution (VWR International Black (0.25 mm or 0.01 in i.d., P/N 052306 for 5 ft) P/N BDH5018) to calibrate the pH meter tubing: liquid line connections from the pump to the injection valve samples Red (0.127 mm or 0.005 in i.d., P/N 052310 for 5 ft) Heparin, sodium salt, grade 1A, porcine intestinal mucosa tubing: liquid line connections from the injection (Sigma-Aldrich P/N H3393) valve to columns and cell. Heparin, sodium salt (Sigma-Aldrich P/N H4784) 10 µL PEEK sample loop (P/N 042949) Chondroitin sulfate A, sodium salt, bovine trachea pH Meter with pH electrode (Sigma-Aldrich P/N 9819) Magnetic stirrer

100 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection conditions PREPARATION OF SOLUTIONS AND REAGENTS Column: IonPac AG11 Guard, 2 × 50 mm Caution: Sodium perchlorate is a strong oxidizer (P/N 044079) and is incompatible with reducing agents and flammable IonPac AS11 Analytical, 2 × 250 mm solvents. Review the Material Safety Data Sheets (MSDS) (P/N 044077) and consult local safety personnel for proper handling, Flow Rate: 0.22 mL/min safety precautions, and waste disposal. Mobile Phase: A: 2.6 mM Sodium phosphate monobasic, pH = 3.0 General Tips When preparing mobile phase solutions, it is B: 1.0 M Sodium perchlorate in 2.6 mM essential to use high quality, Type 1 water, 18.2 MΩ-cm sodium phosphate monobasic, pH = 3.0 resistivity, deionized water. Prepare 1 L of degassed, Gradient: See Table 1 deionized water weekly for the AS flush solution using Column Temp.: 30 °C* vacuum filtration. Standard solutions and Mobile Phase Tray Temp.: 10 °C A were prepared with deionized water (unless otherwise Inj. Volume: 10 µL specified) according to the proposed USP Monograph Detection: UV absorbance at 202 nm described below. Typical System Use glass containers whenever possible for storing and backpressure: 1100–1300 psi preparing mobile phase solutions to prevent UV-detectable leachable compounds from plastic containers that can result Run time: 75 min in increased detector noise and contaminant peaks. *Since the completion of this work, the USP has published a revision that raises the temperature to 40 °C.12 3 M Phosphoric Acid for pH Adjustment Add 1 mL of 85% phosphoric acid to 4 mL of deionized water. Mix thoroughly. This solution will be Table 1: Gradient Conditions used to adjust the pH of both mobile phase solutions. Time Mobile Mobile Dispense 10–100 µL aliquots of 3 M phosphoric acid (min) Phase A Phase B Elution (%) (%) using a micropipettor with a filtered pipette tip to prevent the acid from damaging the micropipettor. 0 80 20 Start linear gradient 60 10 90 End linear gradient. Start the linear gradient to the initial Mobile Phase Solutions conditions Mobile Phase A: 2.6 mM Sodium Phosphate Monobasic 61 80 20 Equilibrate at the initial Dihydrate, pH = 3.0 conditions Prepare 2 L of Mobile Phase A by weighing 75 80 20 End run 0.80 g of sodium phosphate monobasic dihydrate

(NaH2PO4 • 2H2O, FW 155.99) into a 2 L volumetric flask, dissolving it in ~1000 mL of deionized water, and then diluting to the mark with deionized water. Add a stir bar, stir with a magnetic stirrer while adjusting the pH

to 3.0 using incremental 50 µL additions of 3 M H3PO4 to pH 3.2, and then smaller increments (5–20 µL) to pH 3.0.

Typically, 300–350 µL of 3 M H3PO4 is used to adjust a 2 L solution to pH 3.0. Set aside 1 L as the diluent for Mobile Phase B to dissolve the sodium perchlorate reagent. The remaining 1 L will be filtered and degassed after Mobile Phase B is prepared.

101 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection Mobile Phase B: 1.0 M Sodium Perchlorate Heparin Standard Solution

Monohydrate in 2.6 mM NaH2PO4 Dihydrate, pH = 3.0 To prepare a 31 mg/mL heparin standard solution, Prepare 1 L of Mobile Phase B by weighing 140 g dissolve the contents of the USP Heparin Sodium

of sodium perchlorate monohydrate (NaClO4 • H2O, FW Identification RS (9.3 mg) ampoule with 300 µL of 140.46) into a 1 L volumetric flask, and dissolving the deionized water. Gently mix until fully dissolved. Transfer reagent in ~ 500 mL of pH-adjusted Mobile Phase A. The the standard solution into a 300 µL autosampler vial. dissolution of sodium perchlorate is endothermic so allow at least 10 min to equilibrate to room temperature before Standard Solutions diluting to 1 L with pH-adjusted Mobile Phase A. Add a Dermatan Sulfate Standard Solution stir bar and stir with a magnetic stirrer while adjusting the A 20 mg/mL dermatan sulfate standard was prepared pH to 3.0 with 3 M H3PO4. Typically, 5–40 µL is used to by adding 100 µL of the 60 mg/mL dermatan sulfate stock adjust a 1 L solution to pH 3.0. solution and 200 µL of deionized water into a 300 µL Filtering and Degassing Mobile Phase Solutions autosampler vial. Mix the solution thoroughly. Filter and degas each mobile phase separately using vacuum filtration with applied vacuum for 10 min. 1% (v/v) System Suitability Solution Transfer each mobile phase to a separate 1 L glass bottle, To prepare 300 µL of 1% (v/v) of dermatan sulfate immediately cap the bottle, connect it to the corresponding and OSCS in 20 mg/mL heparin sodium standard solution, mobile phase line, and place the mobile phase solutions pipette 3 µL of the 20 mg/mL dermatan sulfate, 10 µL under ~ 4–5 psi of nitrogen or other inert gas. Prime the of the 60 mg/mL System Suitability stock solution, pump with the new solutions and equilibrate the column for 176 µL of the 31 mg/mL heparin standard solution, and 1 h at the starting conditions prior to use. 111 µL deionized water into a 300 µL autosampler vial. Mix thoroughly. This solution can be prepared more easily Standard Stock Solutions when the OSCS RS is available by diluting 3 µL each of Stock solutions were prepared at higher concentra- 20 mg/mL dermatan sulfate and OSCS in 294 µL of tions than proposed by the USP to prepare the 1% (v/v) 20 mg/mL heparin standard solution. This 300 µL volume System Suitability standard. is sufficient for six full loop injections of 10 µL (2x the sample loop volume plus 25 µL used for small loops) or Dermatan Sulfate 15 partial loop injections of 10 µL from 25 µL loop with a To prepare a 60 mg/mL dermatan sulfate stock 5 µL cut volume (2× cut volume plus the injection volume). solution, dissolve 30 mg of chondroitin sulfate B (dermatan sulfate) in 500 µL of deionized water. Sample preparation Heparin System Suitability RS Heparin Samples This stock solution was prepared with USP System To prepare 20 mg/mL heparin samples, first prepare Suitability RS, the only reference standard containing a 60 mg/mL heparin solution by dissolving 30 mg of the OSCS available at the time of our experiments. To (Sigma-Aldrich grade 1A or Sigma) heparin sample in prepare a 60 mg/mL solution of USP System Suitability 500 µL of deionized water, and then gently mixing until stock standard, dissolve 30 mg of the USP System fully dissolved. Add 100 µL of 60 mg/mL heparin to 200 Suitability RS in 500 µL of deionized water. The final µL of deionized water and then transfer the 20 mg/mL solution was expected to contain approximately heparin sample into a 300 µL autosampler vial. 54 mg/mL USP heparin and 6 mg/mL USP OSCS based To prepare 20 mg/mL heparin with 0.2 mg/mL on the composition of >90% heparin and <10% OSCS dermatan sulfate and OSCS samples similar to the 1% listed in the MSDS. This stock solution was used to (v/v) System Suitability standard solution, pipette 3 µL of prepare the 1% (v/v) System Suitability standard and to 20 mg/mL dermatan sulfate, 10 µL of 60 mg/mL System add OSCS to other standards and sample solutions. Suitability stock solution, 273 µL of 20 mg/mL heparin sample, and 14 µL deionized water. Transfer the 1% (v/v) System Suitability solution to a 300 µL sample vial. These heparin samples contain 1.8 mg/mL of USP heparin with the remainder of heparin from the Sigma heparin sample.

102 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection SYSTEM PREPARATION AND SETUP shown as a separate channel without overwriting the Plumbing the Chromatography System original data. Right click on the open chromatogram to To plumb the chromatography system, use black select Arithmetic Combination. Channels A and B are PEEK tubing (See description under Equipment section.) the chromatograms of the sample and the water injection between the pump and injection valve, and red PEEK to be subtracted, respectively. Select UV_Vis_1 for both tubing after the injection valve (in the TC module) and chromatograms. For the most accurate subtraction, select before the flow cell. Install ~ 15 cm (~ 6 in) of black the water sample injection closest to the sample injection. PEEK tubing from the pump to the GM-4 gradient The subtracted chromatogram will be shown in a channel, mixer, and ~ 61 cm (~ 24 in) of black PEEK tubing from typically labeled Sub_UV_Vis_1_water_US_Vis_1 and the gradient mixer to port P (or 2) in System 1 injection will be automatically saved with the data file. valve. Install the sample loop into the injection valve ports labeled L (1 and 4), the AS Autosampler transfer line results and discussion into port S (or 5), and the autosampler waste line into port Separation W (or 6). To install the IonPac AS11 column set, connect Figure 1 demonstrates that 20 mg/mL of USP heparin ~ 28 cm (~ 11 in) of red PEEK tubing into port C (or 3) is eluted from the 2 × 250 mm IonPac AS11 anion followed by guard and the analytical column according to exchange column within 35 min using 2.6 mM NaH2PO4 the product manual.10 Connect ~ 43 cm (~ 17 in) of (pH 3) and 0.2 to 0.9 M NaClO4 (pH 3) from 0 to 60 min red PEEK tubing from the end of the analytical column at 0.22 mL/min. The baseline increases as the gradient to the flow cell inlet. Install a 100 µL sample syringe increases. Heparin is the large broad peak eluting from 20 on the AS Autosampler for a 10 µL injection. Enter to 35 min with a maxima at 25.6 min. The width of this the sample syringe and the sample loop sizes into the peak is attributed to the distribution of chain lengths of AS autosampler module under module setup menu and heparin in the heparin sodium sample and the variation in plumbing configuration. the number of sulfate groups per molecule.

Assembling the UV Semi-Micro PEEK Cell Install the semi-micro PEEK flow cell according to the product manual.11 To reduce noise, install the backpressure line from the flow cell kit on the flow cell outlet and before the waste line. After the mobile phase Column: IonPac AG11, AS11 (2 mm) Flow Rate: 0.22 mL/min Mobile Phase: A: 2.6 mM NaH2PO4 Inj. Volume: 10 µL B: 1 M NaCIO in 2.6 mM NaH PO Detection: UV at 202 nm is flowing through the cell, turn on the cell and allow 4 2 4 Gradient: 20% to 90% B from 0 to 60 min, Sample: 20 mg/mL USP heparin 60 min to warm up the lamp. To prevent salt build-up in 90% B to 20% B from 60 to 61 min, Column Temp.: 30 °C Peak: 1. Heparin 20 mg/mL the cell when the cell is inactive overnight or over the AS Temp.: 10 °C weekend, manually flush the cell with five 1 mL aliquots 120 of deionized water using a 1 mL disposable syringe, as a 1 precaution to prevent damage to the tubing. 90% The PEEK semi-micro flow cell is preferred over a stainless steel cell or standard cell because the PEEK mAu material is inert to the mobile phase solutions and the smaller flow path of this cell is designed for the low flow rates used in this application.

20% Post-Acquisition, Background Subtraction %B: 20% -5 The gradient of sodium perchlorate causes a rise in 0 20 40 60 75 Minutes the baseline of the chromatogram. If desired, one can 26167 subtract the baseline from a water injection by using Figure 1. 20 mg/mL USP heparin identification standard. the Chromeleon post-acquisition function, that is the Arithmetic Combination. The new chromatogram is

103 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection Modifying the Mobile Phase Solutions Column: IonPac AG11, AS11 (2 mm) Flow Rate: 0.22 mL/min

Initially, the analytes eluted 7–10 min earlier than Mobile Phase: A: 2.6 mM NaH2PO4 Inj. Volume: 10 µL

B: 1 M NaCIO4 in 2.6 mM NaH2PO4 Detection: UV at 202 nm expected when using mobile phase solutions prepared Gradient: 20% to 90% B from 0 to 60 min, Sample: A: 0.2 mg/mL Dermatan sulfate according to the proposed Stage 2 revision of the 90% B to 20% B from 60 to 61 min, B: 0.2 mg/mL Dermatan sulfate Column Temp.: 30 °C in 20 mg/mL grade 1A heparin Identification B monograph. Two liters of Mobile Phase AS Temp.: 10 °C Post Acquisition: Baseline subtraction from a A were prepared with 0.80 g of the anhydrous salt of 90 water injection A B NaH PO and two liters of Mobile Phase B were prepared 2 2 4 Peak: 1. Dermatan sulfate 0.2 0.2 mg/mL in the same way with the same amount and type of sodium 2. Heparin — 20 phosphate plus 280 g of the anhydrous salt of NaClO4 (pH 3). To increase the retention of the analytes, we mAu reduced the concentrations of the mobile phase solutions in collaboration with the USP. This was accomplished by adding the same weight of the monohydrated forms 1 of each reagent as listed above without adjusting for the B A differences in formula weight. We also evaluated the use of -15 020406075 the dihydrate salt of NaH2PO4, which effectively reduced Minutes 26168 the molarity from 3.3 to 2.6 mM NaH2PO4 and from 1.1 to

1.0 M NaClO4, resulting in increased retention times and Figure 2. Comparison of 0.2 mg/mL dermatan sulfate A) without resolution of the other analytes from heparin. To further and B) with 20 mg/mL heparin. improve the consistency of the mobile phase preparation, 1 L of pH-adjusted Mobile Phase A from a 2 L preparation was used as the diluent to prepare 1 L of Mobile Phase B. Column: IonPac AG11, AS11 (2 mm) Flow Rate: 0.22 mL/min

Mobile Phase: A: 2.9 mM NaH2PO4 (Sample A) Inj. Volume: 10 µL A: 2.6 mM NaH PO (Sample B) Detection: UV at 202 nm Sample Analysis 2 4 B: 1 M NaCIO4 in Mobile Phase A Samples: A–B: 20 mg/mL Sigma Figure 2 shows the separation of 0.2 mg/mL dermatan Gradient: 20% to 90% B from 0 to 60 min, heparin with 0.2 mg/mL 90% B to 20% B from 60 to 61 min, dermatan sulfate and OSCS sulfate, with and without 20 mg/mL grade 1A heparin. Column Temp.: 30 °C In these chromatograms, we applied the background AS Temp.: 10 °C Peaks: 1. Dermatan sulfate 0.2 mg/mL 100 2. Heparin 20 3. OSCS 0.2 subtraction of a water injection to clearly demonstrate 2 that dermatan sulfate has a smaller peak response when 20 mg/mL heparin is present. This lower response when heparin is present indicates that the heparin may be slightly mAu overloading the column. The resolution of dermatan sulfate (1.1 ± 0.1 USP) in our experiments marginally meets the USP specification of NLT (not less than) 1.0. (In the later 1 3 heparin monograph, the column temperature conditions B were changed to 40 °C to improve resolution of dermatan A 12 -15 sulfate and OSCS from heparin.) These results show that 020406075 Minutes the 20 mg/mL heparin peak starts eluting from the column 26169 within 2 min after the dermatan peak is eluted and further Figure 3. Comparison of 1% System Suitability sample separated implies that dermatan sulfate may be unresolved from with Mobile Phase A prepared with A) monohydrate and B) heparin at >20 mg/mL concentrations. dihydrate salts of sodium phosphate monobasic. Figure 3B shows the separation of 0.2 mg/mL dermatan sulfate and OSCS from 20 mg/mL heparin in the 1% (v/v) System Suitability Standard. The resolution (USP) of heparin from the critical contaminant OSCS is 1.8 ± 0.1, well within the USP specification of NLT 1.5.

104 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection However, the resolution of dermatan sulfate from heparin Method Ruggedness (1.1 ± 0.1) is marginally within the USP specification of Effect of Mobile Phase pH NLT 1.0. The retention times of dermatan sulfate, heparin, Mobile phase solutions prepared with the dihydrate and OSCS were 16.8 ± 0.6, 25.1 ± 0.3, and 41.7 ± 0.2 min salt of NaH2PO4 and monohydrate salt of NaClO4 were (n = 6), respectively. These retention times are adjusted to pH 2.9 and 3.1 to compare to similar solutions 3–8 min earlier than reported by the USP (20, 30, 50 min). adjusted to pH 3.0. These differences in mobile phase pH To determine the elution of other glycosaminoglycans, had very little effect on retention times and a small effect we tested commercially available chondroitin sulfate A at on the resolution (USP) of dermatan sulfate (0.93 ± 0.06) 0.2 mg/mL in 20 mg/mL heparin (not shown). We found and OSCS (1.68 ± 0.08) from heparin. that chondroitin sulfate A elutes at the same time as dermatan sulfate. Column Reproducibility Two IonPac AS11 columns were compared from the Precision and Reproducibility same lot using mobile phase solutions prepared from the To evaluate the method precision, we determined the monohydrate salts and performing triplicate injections heparin peak area and retention time RSDs of dermatan of 0.2 mg/mL dermatan sulfate in 20 mg/mL grade 1A sulfate, OSCS, and heparin with triplicate injections heparin. We found negligible differences in the retention of 1% (v/v) System Suitability samples containing times of dermatan sulfate (<0.7 min) and heparin 0.2 mg/mL dermatan sulfate, 0.2 mg/mL OSCS, and (<1.2 min) and no change in resolution of dermatan 20 mg/mL heparin over three days with mobile phase sulfate from heparin (1.1 ± 0.1 USP). solutions prepared daily. During this experiment, we used

the monohydrate salt of NaH2PO4. During this period of Method Stability time, we evaluated the effect of the salt hydration form The IonPac AS11 column life was evaluated for on resolution and retention times (Figure 3). The mobile this method by performing 310 injections of heparin phase solutions in these 3-day experiments were prepared over 27 consecutive days. Injections of the 1% System with 0.8 g of of NaH2PO4 per 2 L without adjusting for Suitability standards were interspersed on average every formula weight (2.9 mM NaH2PO4). 15 injections, to monitor the change in the resolution The retention times of the analytes were slightly of dermatan sulfate and OSCS from heparin. This lower with this mobile phase than with the mobile experiment showed that the IonPac AS11 column was phase prepared with the dihydrate salt. The intra-day not affected by the 310 heparin injections (Figure 4). and between-day heparin peak area RSDs averaged The resolution of dermatan sulfate from heparin at the 0.67 ± 0.12 and 0.60 respectively, well within the USP start of the experiment was 1.02 ± 0.07 (n = 3), which specification of NMT (not more than) 2. The intra-day was similar to the resolution at the end of the experiment retention time RSDs of dermatan sulfate, heparin, and (1.09 ± 0.06, n = 3). The resolution averaged 1.04 ± 0.07 OSCS were <1.2 for all three peaks, whereas the between- (n = 25) for the experiment, marginally exceeding the day RSDs were 3.6, 1.1, and 0.5 respectively. Although USP specification (NLT 1). The resolution of heparin

the monohydrate salt of NaH2PO4 was used in this from OSCS was also similar at the beginning and end of experiment, we recommend the dihydrate salt of NaH2PO4 the experiment with resolution values of 1.74 ± 0.07 for mobile phase preparation (used for all other work in (n = 3) and 1.84 ± 0.06 (n = 3), respectively. The average this AN except the column study, including the method resolution was 1.76 ± 0.12 (n = 25). The retention time ruggedness studies that follow). We did not repeat these of heparin averaged 25.43 ± 0.52 min (2.0% RSD; experiments with using the dihydrated form of the salt to n = 310). The changes in mobile phase solutions had prepare the mobile phase. minimal effect on the retention times. Communication with a customer since the completion of this work indicates that a column temperature of 40 °C increases both resolution factors.

105 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection 45 REFERENCES 40 1. Goldstein, J. Making Heparin is a Dirty Job. The

(min) 35 Mobile Phase Change WSJ. February 21, 2008. 30 2. Tyrrell, D.; Horne, A. P.; Holme, K. R.; Preuss, J. M. 25 H.; Page, C. P. Heparin in Inflammation: Potential 20 Retention time 15 Therapeutic Applications Beyond Anticoagulation. 10 Adv. Pharmacol. 1999, 46, 151–162. 050 100 150 200 250 300 350 Injections (#) 3. Rutty, C. J., Ph.D. Miracle Blood Lubricant: Heparin Dermatan sulfate OSCS Connaught and the Story of Heparin Contact 26170 (Employee News of Connaught Laboratories Ltd.) Figure 4. Retention time stability of 20 mg/mL heparin, 1996, 9 (4), 1928–1937. 0.2 mg/mL dermatan sulfate, and 0.2 mg/mL OSCS over 27 days. 4. FDA Press Release, Baxter’s Multiple-Dose Vial Heparin Linked to Severe Allergic Reactions. Federal CONCLUSION Drug Administration. February 11, 2008. In this AN, 1% (v/v) dermatan sulfate and OSCS 5. Guerrini, M.; Beccati, D.; Shriver, Z.; Naggi, A.; were separated from heparin using the USP Identification Viswanathan, K.; Bisio, A.; Capila, I.; Lansing, J. B method, Stage 2 revision of the Heparin Sodium C.; Guglieri, S.; Fraser, B.; Al-Hakim, A.; Gunay, monograph. This application takes advantage of the N. S.; Zhang, Z.; Robinson, L.; Buhse, L.; Nasr, unique properties of the IonPac AS11 column to resolve M.; Woodcock, J.; Langer, R.; Venkataraman, G.; dermatan sulfate and the critical OSCS contaminant from Linhardt, R. J.; Casu, B.; Torri, G.; Sasisekharan, R. heparin, thereby allowing OSCS-contaminated heparin Oversulfated Chondroitin Sulfate is a Contaminant to be detected and quarantined from worldwide supplies. in Heparin Associated with Adverse Clinical Events. Additionally, the 2 mm column format provides the Nat. Biotechnol. 2008, 26, 669–675. advantage of operating at low flow rates, which reduces the 6. Heparin Sodium monograph in The United States mobile phase consumption and waste. Pharmacopeia 31, Supplement 2, NF 26; American Pharmaceutical Association, Washington, DC, 2008. PRECAUTIONS 7. Heparin Sodium monograph in The United States Consistent mobile phase preparation is critical to Pharmacopeia 32, Supplement 2, NF 25; American consistent retention times and separation. Do not prepare Pharmaceutical Association, Washington, DC, 2009. >20 mg/mL concentrations of heparin in working solutions 8. Heparin Sodium, Pharmacopeial Forum 2009, or standards, as dermatan sulfate may not be resolved. 35 (2), 1–10. 9. Dionex Corporation. Determination of Organic SUPPLIERS Impurities in Heparin by HPAE-PAD using CarboPac Sigma-Aldrich, Inc., P.O. Box 951524, Dallas, TX PA20 Column, Application Note 233; LPN 2286. 75395-1524, 1-800-325-3010 Sunnyvale, CA, 2009. www.sigmaaldrich.com 10. Dionex Corporation. Product Manual for IonPac U.S. Pharmacopeia, 12601 Twinbrook Parkway, AG11 Guard and AS11 Analytical Column; Document Rockville, MD, USA 20852–1790 No. 034791. Sunnyvale, CA, 2006. 1-800-227-8772 www.usp.org 11. Dionex Corporation. Product Manual for ICS Series Variable Wavelength Detector; Document No. VWR International, Inc., Goshen Corporate Park West, 065141. Sunnyvale, CA, 2006. 1310 Goshen Parkway, West Chester, PA 19380 12. Heparin Sodium, Pharmacopeial Forum 2009, 1-800-932-5000 www.vwrsp.com 35 (5), 1–4.

106 Determination of Oversulfated Chondroitin Sulfate and Dermatan Sulfate in Heparin Sodium Using Anion-Exchange Chromatography with UV Detection Application Update 178

A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium

INTRODUCTION While this method is able to achieve the objective Heparin is a complex highly sulfated of resolving the target compounds from heparin, the glycosaminoglycan that has been used since the early resolution between DS and heparin often just meets the 20th century as an anticoagulant for the treatment of USP specification of not less than (NLT) 1.0. In addition, thrombosis.1 It is estimated that more than 10 million the analysis run time is 75 min, and using sodium Americans receive heparin annually and more than perchlorate as the eluent raises environmental and human 70 million vials are sold for cardiac surgery, dialysis, health concerns. Therefore, there is significant opportunity and a variety of other uses. In 2008, researchers detected to improve this analytical method for the determination of a contaminant now known as oversulfated chondroitin DS and OSCS in heparin. sulfate (OSCS) in heparin samples that was associated Alternative chromatographic methods have been with life-threatening allergic reactions in hundreds of investigated to improve the current USP method for the patients. The samples were also determined to contain determination of heparin. For example, weak anion- the impurity dermatan sulfate (DS), which is commonly exchange (WAX) stationary phases were evaluated to found in heparin due to incomplete purification.2 provide alternative columns to the USP method, which To address the immediate health concerns, nuclear specifies a strong anion-exchange column. Although one magnetic resonance (NMR) and capillary electrophoresis publication demonstrated improvement in the method by (CE) methods were rapidly developed to detect OSCS in increasing sample throughput by significantly decreasing the suspected heparin samples.3 Although CE was able run times, the separation still required perchlorate to to detect the contaminant, OSCS and DS were not fully resolve DS and OSCS from heparin.7 Larive et al. also resolved from heparin. Due to these and other inherent demonstrated that a WAX column could be used to challenges of implementing CE in a quality control resolve these compounds.8 In this example, a mildly environment, the United States Pharmacopeia (USP) alkaline mobile phase was used, but the sensitivity of published a chromatographic method that resolved DS OSCS was inferior to the current USP method. and OSCS from heparin.4 The current USP Heparin In 2009, researchers at the United States Food and Sodium monograph contains a chromatographic identity Drug Administration (US FDA) published a method using method (Identification B) that prescribes the separation of the IonPac AS11-HC column with a sodium chloride DS and OSCS from heparin using a USP type L61 column in Tris gradient to resolve the target compounds from (IonPac® AS11) with a sodium perchlorate gradient heparin.9 This method provides several benefits over the followed by absorbance detection at 202 nm.5 Additional current USP method, such as eliminating the need for information on this method can be found in Dionex perchlorate in the mobile phase, reducing the analysis Application Note 235.6 time from 75 min to 40 min, and improving the resolution between DS and heparin.

107 A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium The testing here demonstrates an improved method REAGENTS AND STANDARDS for resolving DS and OSCS in heparin with an IonPac Deionized (DI) water, Type 1 reagent-grade, 18.2 MΩ-cm AS11-HC (2 × 250 mm) column using a NaCl gradient resistivity followed by absorbance detection at 215 nm, which is Use only ACS reagent-grade chemicals for all reagents similar to the US FDA method previously described. The and standards higher capacity of the IonPac AS11-HC column, relative Heparin, Grade 1A from porcine mucosa (Sigma-Aldrich to the IonPac AS11 column described in the current USP P/N H-3393) method, minimizes the possibility of column overload and Heparin Sodium Identification Standard (9.3 mg, USP provides improved resolution between DS and heparin. P/N 1304038) The microbore column format was chosen to reduce eluent consumption and thereby reduce the time and Heparin Sodium System Suitability Standard labor required for manual eluent preparation. This study (>90% heparin, <10% oversulfated chondroitin demonstrates the method’s ruggedness and precision, and sulfate, 50 mg, USP P/N 1304049) the ability to detect <1% (w/w) DS and OSCS in heparin. Chondroitin sulfate B (dermatan sulfate, sodium salt), sodium salt from porcine intestinal mucosa, EQUIPMENT >90% lyophilized powder (Sigma-Aldrich Dionex ICS-3000 systema including: P/N C3788) SP Single Pump module with a GM-4 gradient mixer Sodium chloride (FW 58.44, VWR International TC Thermal Compartment with a 6-port injection P/N JT3624-19) valve Tris hydrochloride (Tris[hydroxymethyl]aminomethane VWD Variable Wavelength Detector 3400 with hydrochloride, FW 157.1, VWR International PEEK™ semi-micro flow cell (2.5 µL, 7 mm, P/N EM1.08219.9025) (P/N 6074-0300) pH 7 buffer (VWR International P/N BDH5046) AS Autosampler with Sample Tray Temperature pH 4 buffer (VWR International P/N BDH5018) Controlling option, 100 µL sample syringe Phosphoric acid, 85–87% (H3PO4, VWR International (P/N 055064), and 1.5 mL sample tray P/N JT0260) Chromeleon® 7.1 Chromatography Data System (CDS) software CONDITIONS Vial Kit, 1.5 mL glass with caps and septa (P/N 055427) Column: IonPac AS11-HC Guard, 2 × 50 mm or 0.3 mL polypropylene sample vials with caps and (P/N 052963) slit septa (P/N 055428) IonPac AS11-HC Analytical, Nalgene® Media-Plus with 90 mm, 0.45 µm nylon 2 × 250 mm (P/N 052961) filter (Nalge Nunc International P/N 164-0020) or Mobile Phases: A: DI water equivalent nylon filter B: 2.5 M Sodium chloride, Vacuum pump 20 mM Tris (pH 3) 10 µL PEEK sample loop (P/N 042949) Gradient: See Table 1. pH Meter with pH electrode Flow Rate: 0.20 mL/min Magnetic stirrer Column Temp.: 40 °C Sample Volume: 10 µL aThis application can be performed on a Dionex ICS-5000 system. Detection: UV, 215 nm Background: 0–15 mAU over the gradient Noise: <100 µAU System Backpressure: 1550 psi Run Time: 40 min

108 A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium Table 1. Gradient Conditions Mobile Phase B: 2.5 M Sodium Chloride, 20 mM Tris (pH = 3) Time Mobile Phase A Mobile Phase B (min) (%) (%) Prepare 1 L of Mobile Phase B by weighing 146.1 g 0 95 5 of sodium chloride and 3.14 g of Tris hydrochloride 2 95 5 (FW 157.1) into a 1 L volumetric flask and dissolve the reagent in ~800 mL of deionized water. Stir with a 26 5 95 magnetic stirrer until the reagents are fully dissolved. 31 5 95 Remove the stir bar, dilute to the 1 L mark, and mix by 32 95 5 inverting the flask. Carefully re-add the stir bar and stir 40 95 5 while adjusting to pH 3.0 with 3 M phosphoric acid. Typically, 5–40 µL is used to adjust a 1 L solution to pH 3.0. Filter and degas using vacuum filtration with PREPARATION OF SOLUTIONS AND REAGENTS applied vacuum for 10 min. Transfer the mobile phase General Tips to a separate 2 L glass bottle, immediately cap the bottle, Use high quality, Type 1, 18.2 MΩ-cm resistivity DI connect it to the corresponding mobile phase line, and water to prepare mobile phase solutions and standards. place the mobile phase solutions under ~4–5 psi of Prepare 1 L of degassed DI water weekly for the nitrogen or other inert gas. Prime the pump with the new AS Autosampler flush solution using vacuum filtration. mobile phase solution and equilibrate the column for 1 h Use glass containers for storing and preparing mobile at the starting conditions prior to use. phase solutions to minimize leachable compounds from plastic containers that can result in increased Standard Solutions chromatography noise and contaminant peaks. Prepare DS, Heparin System Suitability RS, heparin stock standard, and 1% (v/v) System Suitability working 3 M Phosphoric Acid for pH Adjustment solutions according to the instructions described in Add 1 mL of 85% phosphoric acid to 4 mL DI AN 235.5 water and mix thoroughly. This solution will be used Note: A 300 µL volume is sufficient for 6 full-loop to adjust the pH of Mobile Phase B. Dispense 500 µL injections of 10 µL (2 × the sample loop volume plus aliquots of 3 M phosphoric acid using a micropipettor 25 µL used for small loops) or 15 partial-loop injections with a filtered pipette tip to prevent the acid from of 10 µL from a 25 µL loop with a 5 µL cut volume damaging the micropipettor. (2 × cut volume plus the injection volume).

Mobile Phase Solutions SYSTEM PREPARATION AND CONFIGURATION Mobile Phase A: Deionized Water To configure the system, follow the instructions in Degas 2 L of DI water. Transfer the DI water to a AN 235 and the system and column product manuals. 2 L eluent bottle, connect the bottle to the Mobile Use red PEEK (0.005 in i.d.; 0.013 mm i.d.) tubing for Phase A line, and place the mobile phase under 5 psi all liquid lines from the injection valve to the detector. of inert gas, such as nitrogen. Prime the pump with the Assemble the UV semi-micro PEEK cell according to the new mobile phase. instructions in AN 235. Note: The semi-micro PEEK cell was selected for the short flow path and suitability of the PEEK material for the acidic high-salt mobile phases used here.

109 A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium RESULTS AND DISCUSSION Column: IonPac AG11-HC, AS11-HC (2 mm) In this study, the separation of DS from heparin Mobile Phases: A: Degassed deionized water B: 2.5 M NaCl, 20 mM TRIS (pH 3) was evaluated using similar conditions published in the Gradient: 5% B from 0–2 min, 5 to 95% B from literature7-9 with the objectives of improving the resolution 2–26 min, 95% B from 26–31 min, 95 to 5% B from 31–32 min, of DS from heparin, avoiding a perchlorate-based 5% B from 32–40 min Column Temp.: 40 °C mobile phase, and reducing the analysis time. The initial Flow Rate: 0.20 mL/min investigation evaluated the feasibility of using the Inj. Volume: 10 µL Detection: UV, 215 nm ® ® ProPac WAX-10 column, the ProSwift WAX-1S Standard: 1% DS and OSCS in 20 mg/mL column, and the ProSwift SAX-1S monolith column for Grade 1A heparin Peaks: 1. Dermatan Sulfate 0.2 mg/mL this application. The target analytes were separated using 2. Heparin 20 either conditions similar to those described in the current 3. OSCS 0.2

USP monograph, a NaCl gradient in phosphate buffer 115 2 at pH 3, or a NaCl gradient in phosphate buffer at pH 6–10 with absorbance detection at 202 or 215 nm. For the ProPac WAX-10 column, the separation of DS and heparin required perchlorate to elute the analytes from the column. In addition, the resolution between these 3 compounds did not meet the USP specification of NLT mAU 1 1. At higher mobile phase pH (6–10), the compounds were not retained on the column and therefore eluted at or near the void. For the ProSwift WAX-1S and SAX-1S columns, DS and heparin could not be resolved using a NaCl mobile phase with either Tris or phosphate at pH 3. Researchers at the US FDA demonstrated that the –35 01020430 0 IonPac AS11-HC column could separate DS and OSCS Minutes 28089 from heparin using a NaCl gradient in Tris at pH 3 that met the objectives described above.9 Figure 1. Heparin standard separation prior to Therefore, further baseline subtraction. investigation and minor modifications to this approach were made to produce an improved method for deter- Using the conditions described in this study, the mining the target compounds in heparin sodium. The resolution between DS and heparin was 1.9, which modifications included using a 2 mm IonPac AS11-HC is nearly double the USP specification and therefore column and increasing the column temperature from 35 to a significant improvement relative to the current o 40 C, which produced slightly better results. monograph. All analytes eluted from the column in Figure 1 demonstrates the separation of 0.2 mg/mL about 40 min, which is nearly half the time required DS and OSCS from 20 mg/mL heparin. To displace for the current USP method. To correct for the increase heparin from the column, a significantly higher salt in absorbance during the gradient, the baseline concentration (2.5 M NaCl) is required relative to the was subtracted from a DI water injection using the perchlorate concentration in the current USP method. Chromeleon CDS processing method (Figure 2). The However, this was expected due to the weaker eluting baseline subtraction improves peak integration and effect of chloride relative to perchlorate. In addition, therefore the method’s accuracy and precision. the IonPac AS11-HC column has more than six times the capacity of the IonPac AS11column and, therefore, requires greater eluent strength.

110 A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium Column: IonPac AG11-HC, AS11-HC (2 mm) 3 Rs = 2.07 ± 0.07 Mobile Phases: A: Degassed deionized water 2.5 B: 2.5 M NaCl, 20 mM TRIS (pH 3) Gradient: 5% B from 0–2 min, 5 to 95% B from 2 2–26 min, 95% B from 26–31 min, 1.5 95 to 5% B from 31–32 min, 5% B from 32–40 min Rs = 1.93 ± 0.05 Column Temp.: 40 °C 1

Flow Rate: 0.20 mL/min Resolution (USP) 0.5 Inj. Volume: 10 µL Detection: UV, 215 nm 0 Data Handling: Baseline subtraction, water injection 0102030405060708090 100 Peaks: 1. Dermatan Sulfate 0.2 mg/mL Injection Number 2. Heparin 20 Dermatan Sulfate OSCS 3. OSCS 0.2 95.0 28087 90

2 Figure 3. Resolution reproducibility over three days.

The between-day retention time and peak area precisions (RSDs) for the three compounds were

mAU <0.5 and ≤6.0, respectively. As shown in Table 2, the peak area precision for heparin was <0.6, which is well 5.0 within the USP specification of NMT 2. The resolution 5.0% B between DS and heparin had an average RS value of 2.0 3 1 ± 0.07 and the resolution between heparin and OSCS

had an average RS value of 1.93 ± 0.05 over the three-day –10 study. In comparison, the DS/heparin and heparin/OSCS 01020430 0 resolutions using the conditions described in the current Minutes 28086 USP method were determined to be 1.04 ± 0.07 and 1.76 ± 0.12, respectively in AN 235. Therefore, this method Figure 2. Separation of 1% dermatan sulfate and OSCS in 20 mg/mL heparin on an IonPac AS11-HC column using a significantly improves the resolution between these salt in Tris gradient, UV detection, and baseline subtraction. compounds, while avoiding sodium perchlorate in the mobile phase. Figure 3 demonstrates the stability of the resolution The method performance was evaluated by between DS and OSCS from heparin for 82 injections determining the between-day precision from over three days, which indicates no loss in column 82 injections over three days, the effect of daily mobile capacity. Column temperatures of 35 and 40 °C were also phase preparation, and differences between a column evaluated as part of this study. Although the results for the temperature of 35 and 40 °C. In addition, two different target compounds were not significant between the two IonPac AS11-HC column lots were investigated to temperatures, a column temperature of 40 °C provided evaluate any potential variability in the separation. a slight improvement in peak response for OSCS and therefore was used for this method.

Table 2. Summary of Reproducibility Experiments Over Three Daysa

Analyte Retention Time (min) RSD Peak Area (mAU-min) RSD Resolution from Heparin (Rs) RSD Dermatan Sulfate 20.3 ± 0.08 0.37 1.59 ± 0.10 6.0 2.04 ± 0.07c 3.7 Heparin 24.6 ± 0.04 0.17 119.3 ± 0.65 0.55b OSCS 27.9 ± 0.12 0.43 4.14 ± 0.17 4.1 1.93 ± 0.05d 2.5 an = 82 bUSP requirement is NMT 2 for n = 3 cUSP requirement is NLT 1 dUSP requirement is NLT 1.5

111 A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium In comparing the separation between two different REFERENCES columns lots, the retention times, peak area responses, 1. Guerrini, M.; Zhang, Z.; Shriver, Z.; Naggi, A.; and resolutions were nearly identical. These results Masuko, S.; Langer, R.; Casu, B.; Linhardt, R.J.; demonstrate the method robustness and therefore the Torri, G.; Sasisekharan, R. Orthogonal Analytical ability of quality control laboratories to use this method Approaches to Detect Potential Contaminants in for routine analysis. Heparin. PNAS, 2009, 106 (40), 16956–16961. 2. Usdin, S. The Heparin Story. Int. J. Risk & Safety in CONCLUSION Medicine 2009, 21, 93–103. This study demonstrates an improved separation of 3. United States Pharmacopeia and the National 1% DS and OSCS from heparin using an IonPac Formulary, Heparin Sodium. USP 32, NF 27, 2009. AS11-HC column with a NaCl/Tris (pH 3) mobile 4. Heparin Sodium, Pharmacopeial Forum 2009, 35 (2), phase and UV absorbance detection at 215 nm. The 1–10. method takes advantage of the high capacity and strong 5. United States Pharmacopeia and the National anion-exchange properties of the column to improve the Formulary, Heparin Sodium. USP 34, NF 29, 2011. resolution of the two contaminants in heparin. In addition, 6. Dionex Corporation, Determination of Oversulfated the method eliminates the need for sodium perchlorate Chondroitin Sulfate and Dermatan Sulfate in Heparin in the mobile phase, but still reduces the analysis time Sodium Using Anion-Exchange Chromatography nearly 50% compared to the current USP method. This with UV Detection. Application Note 235; LPN 2306, study also demonstrates good column stability based on Sunnyvale, CA, 2010. consistent resolution between heparin and the critical 7. Hashii, N.; Kawasaki, N.; Itoh, S.; Qin, Y.; Fujita, contaminants over three days. In addition, the higher N.; Hattori, T.; Miyata, K.; Bando, A.; Sekimoto, Y.; capacity of the IonPac AS11-HC column relative to the Hama, T.; Kashimura, M.; Tatsumi, M.; Mabuchi, IonPac AS11 column reduces the possibility of column K.; Namekawa, H.; Sakai, T.; Hirose, M.; Dobashi, overload and should improve the detection of DS and S.; Shimahashi, H.; Koyama, S.; Herr, S. O.; Kawai, OSCS in heparin. This study provides a faster more K.; Yoden, H.; Yamaguchi, T. Heparin Identification sensitive method to detect OSCS and DS contamination Test and Purity Test for OSCS in Heparin Sodium and in the worldwide heparin supply, thus preventing future Heparin Calcium by Weak Anion-Exchange High- OSCS-related fatalities. Performance Liquid Chromatography. Biologicals 2010, 38 (5), 539–543. SUPPLIERS 8. Limtiaco, J.F.K.; Jones, C. J.; Larive, C. K. Sigma-Aldrich, Inc., P.O. Box 951524, Dallas, TX Characterization of Heparin Impurities with HPLC- 75395-1524, U.S.A. Tel: 1-800-325-3010. NMR Using Weak Anion-Exchange Chromatography. www.sigmaaldrich.com Anal. Chem. 2009, 81 (24), 10116–10123. VWR International, Inc., Goshen Corporate Park West, 9. Trehy, M. L.; Reepmeyr, J. C.; Kolinski, R. E.; 1310 Goshen Parkway, West Chester, PA 19380, Westenberger, B. J.; Buhse, L. F. Analysis of U.S.A. Tel: 1-800-932-5000. www.vwrsp.com Heparin Sodium by SAX/HPLC for Contaminants and Impurities. J. Pharm. Biomed. Anal. 2009, 49, 670–673.

112 A Faster Solution with Increased Resolution for Determining Chromatographic Identity and Absence of OSCS in Heparin Sodium Column Selection Guide Pharmaceutical Applications Notebook Thermo Scientific Acclaim Column Selection Guide Please refer to www.thermoscientific.com/dionex for more information

Reversed-Phase (RP) Mixed-Mode HILIC Application-Specific ) A2 1 1 1 A) X-

WA Example Applications 1 8 8 nity P1 i Tr Acclaim Mixed-Mode HILIC- Acclaim HILIC-10 Acclaim Organic Acid Acclaim Surfactant Acclaim Explosives E1 Acclaim Explosives E2 Acclaim Carbamate Acclaim 120 C1 Acclaim 120 C8 Acclaim 300 C1 Acclaim Polar Advantage (P Acclaim Polar Advantage II (P Acclaim Phenyl- Acclaim Acclaim Mixed-Mode Acclaim Mixed-Mode WCX-

High hydrophobicity Fat-soluble vitamins, PAHs, glycerides Neutral Intermediate hydrophobicity Steroids, phthalates, phenolics Molecules Low hydrophobicity Acetaminophen, urea, polyethylene glycols High hydrophobicity NSAIDs, phospholipids Anionic Intermediate hydrophobicity Asprin, alkyl acids, aromatic acids Molecules

s Low hydrophobicity Small organic acids, e.g. acetic acids High hydrophobicity Antidepressants Cationic Intermediate hydrophobicity Beta blockers, benzidines, alkaloids Molecules Low hydrophobicity Antacids, pseudoephedrine, amino sugars

Amphoteric/ High hydrophobicity Phospholipids

General Application Zwitterionic Intermediate hydrophobicity Amphoteric surfactants, peptides Molecules Low hydrophobicity Amino acids, aspartame, small peptides Neutrals and acids Artificial sweeteners Mixtures of Neutral, Anionic, Neutrals and bases Cough syrup Cationic Acids and bases Drug active ingredient with counterion Molecules Neutrals, acids, and bases Combination pain relievers Anionic SDS, LAS, laureth sulfates Cationic Quats, benzylalkonium in medicines Nonionic Triton X-100 in washing tank Surfactants Amphoteric Cocoamidopropyl betaine Hydrotropes Xylenesulfonates in handsoap Surfactant blends Noionic and anionic surfactants Hydrophobic Aromatic acids, fatty acids Organic Acids Hydrophilic Organic acids in soft drinks, pharmaceuticals Explosives U.S. EPA Method 8330, 8330B Carbonyl compounds U.S. EPA 1667, 555, OT-11; CA CARB 1004 Phenols Compounds regulated by U.S. EPA 604 Chlorinated/Phenoxy acids U.S. EPA Method 555 Triazines Compounds regulated by U.S. EPA 619 Environmental Nitrosamines Compounds regulated by U.S. EPA 8270 Contaminants Benzidines U.S. EPA Method 605 Specific Applications Perfluorinated acids Dionex TN73 Microcystins ISO 20179 Isocyanates U.S. OSHA Methods 42, 47 Carbamate insecticides U.S. EPA Method 531.2 Water-soluble vitamins Vitamins in dietary supplements Vitamins Fat-soluble vitamins Vitamin pills Anions Inorgaic anions and organic acids in drugs Pharmacutical Cations Inorgaic cations and organic bases in drugs Counterions Mixture of Anions and Cations Screening of pharmaceutical counterions API and counterions Naproxen Na+ salt, metformin Cl-salt, etc.

114 Column Selection Guide Transferring HPLC Methods to UHPLC Pharmaceutical Applications Notebook Technical Note 75

Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator

Introduction collection rates up to 100 Hz (even when acquiring The goal of every chromatographic optimization is UV-Vis spectra), sets the standard for UHPLC a method that sufficiently resolves all peaks of interest performance. Acclaim® RSLC columns with a 2.2 µm in as short a time as possible. The evolution of packing particle size complete the RSLC dimension. materials and instrument performance has extended A successful transfer from an HPLC method chromatographic separations to new limits: ultrahigh- to an RSLC method requires recalculation of performance liquid chromatography (UHPLC). the chromatographic parameters. Underlying The new Dionex UltiMate® 3000 Rapid Separation chromatographic principles have to be considered to find LC (RSLC) system is ideal for ultrafast, high-resolution the appropriate parameters for a method transfer. With LC. The RSLC system was designed for ultrafast the Method Speed-up Calculator, Dionex offers an separations with flow rates up to 5 mL/min at pressures electronic tool that streamlines the process of optimum up to 800 bar (11,600 psi) for the entire flow-rate range. method transfer. This technical note describes the This industry-leading flow-pressure footprint ensures theory behind the Method Speed-Up Calculator and the highest flexibility possible; from conventional to the application of this interactive, multi-language tool, ultrahigh-resolution to ultrahigh-speed methods. The illustrated with an exemplary method transfer from a RSLC system, with autosampler cycle times of only conventional LC separation to an RSLC separation. You 15 seconds, oven temperatures up to 110 °C, and data may obtain a copy of this calculator from your Dionex representative.

116 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Method Speed-Up Strategy in determining the optimum mobile phase flow rate for The purpose of method speed-up is to achieve highest column efficiency with lowest plate heights. A sufficient resolution in the shortest possible time. simplification of the Van Deemter equation, according to The strategy is to maintain the resolving power of Halász1 (Formula 3), describes the relationship between the application by using shorter columns packed with column efficiency (expressed in plate height H), particle

smaller particles. The theory for this approach is based size dp (in µm) and velocity of mobile phase u (in mm/s): on chromatographic mechanisms, found in almost every chromatography text book. The following fundamental 2 6 p ⋅ ud Formula 3: =2 ⋅ dH ++ chromatographic equations are applied by the Method p u 20 Speed-Up Calculator for the method transfer from conventional to ultrafast methods. The plots of plate height H against velocity u The separation efficiency of a method is stated by the depending on the particle sizes dp of the stationary phase peak capacity P, which describes the number of peaks that (see Figure 1, top) demonstrate visually the key function can be resolved in a given time period. The peak capacity of small particle sizes in the method speed-up strategy: is defined by the run time divided by the average peak The smaller the particles, the smaller the plate height and width. Hence, a small peak width is essential for a fast therefore the better the separation efficiency. An efficiency method with high separation efficiency. The peak width equivalent to larger particle columns can be achieved by is proportional to the inverse square root of the number of using shorter columns and therefore shorter run times. theoretical plates N generated by the column. Taking into Another benefit with use of smaller particles is shown account the length of the column, its efficiency can also be for the 2 µm particles in Figure 1: Due to improved mass expressed by the height equivalent to a theoretical plate H. transfer with small particle packings, further acceleration The relationship between plate height H and plate number of mobile phases beyond the optimal flow rate with N of a column with the length L is given by Formula 1. minimal change in the plate height is possible. Optimum flow rates and minimum achievable plate L Formula 1: N= heights can be calculated by setting the first derivative of H the Halász equation to zero. The optimal linear velocity (in mm/s) is then calculated by Formula 4. Low height equivalents will therefore generate a high number of theoretical plates, and hence small peak width for high peak capacity is gained. Which factors define H? B 10.95 Formula 4: uopt == For an answer, the processes inside the column have to dC p be considered, which are expressed by the Van Deemter equation (Formula 2). The minimum achievable plate height as a function of particle size is calculated by insertion of Formula 4 in Formula 3, resulting in Formula 5. B Formula 2: AH ++= ⋅ uC u

Formula 5: min 3⋅ dH p The Eddy diffusion A describes the mobile phase movement along different random paths through the Chromatographers typically prefer resolution over stationary phase, resulting in broadening of the analyte theoretical plates as a measure of the separation quality. band. The longitudinal diffusion of the analyte against the The achievable resolution R of a method is directly flow rate is expressed by the term B. Term C describes proportional to the square root of the theoretical plate the resistance of the analyte to mass transfer into the number as can be seen in Formula 6. k is the retention pores of the stationary phase. This results in higher band factor of the analyte and k the selectivity. broadening with increasing velocity of the mobile phase. The well-known Van Deemter plots of plate height H 1 k α −1 Formula 6: NR 2 against the linear velocity of the mobile phase are useful ⋅⋅= ⋅ 14 + k 2 α

117 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator If the column length is kept constant and the particle When transferring a gradient method, the scaling size is decreased, the resolution of the analytes improves. of the gradient profile to the new column format and Figure 1, bottom, demonstrates this effect using 5 µm and flow rate has to be considered to maintain the separation 2 µm particles. performance. The theoretical background was introduced by L. Snyder2 and is known as the gradient volume principle. The gradient volume is defined as the mobile

100 phase volume that flows through the column at a defined

gradient time tG. Analytes are considered to elute at constant eluent composition. Keeping the ratio between 10 µm particles the gradient volume and the column volume constant 5 µm particles therefore results in a correct gradient transfer to a different 3 µm particles 2 µm particles column format. Taking into account the changed flow rates F H [µm] and column volume (with diameter dc and length L),

the gradient time intervals tG of the new methods are uopt. 5 µm uopt. 2 µm calculated with Formula 7.

Hmin. 5 µm 2 H Fold Lnew d ,newc min. 2 µm • • Formula 7: , = tt ,oldGnewG • 0 F L ( d ( 015 0 new old ,oldc Linear Velocity u [mm/s] An easy transfer of method parameters can be 700 Separation on 5 µm material achieved by using the Dionex Method Speed-Up Calculator (Figure 2), which incorporates all the mAU overwhelming theory and makes manual calculations unnecessary. This technical note describes the easy method transfer of an example separation applying 0 1.7 2.0 Minutes 3.0 the calculator. Just some prerequisites described in the following section have to be taken into account. 700 Separation on 2 µm material Prerequisites mAU The Method Speed-Up Calculator is a universal tool and not specific for Dionex products. Nevertheless, some prerequisites have to be considered for a successful 0 1.7 2.0 Minutes 3.0 method transfer, which is demonstrated in this technical 25716 note by the separation of seven soft drink additives. Figure 1. Smaller particles provide more theoretical plates and more resolution, demonstrated by the improved separation of three peaks (bottom) and smaller minimum plate heights H in the Van Deemter plot (top). At linear velocities higher than uopt, H increases more slowly when using smaller particles, allowing higher flow rates and therefore faster separations while keeping separation efficiency almost constant. The speed-up potential of small particles is revealed by the Van Deemter plots (top) of plate height H against linear velocity u of mobile phase: Reduc- ing the particle size allows higher flow rates and shorter columns because of the decreased minimum plate height and increased optimum velocity. Consequently, smaller peak width and improved resolution are the result (bottom).

118 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Figure 2. The Dionex Method Speed-Up Calculator transfers a conventional (current) HPLC method to a new (planned) RSLC method.

Column Dimension Table 1. Theoretical Plates Depending on First, the transfer of a conventional method to an Column Length and Particle Diameter RSLC method requires the selection of an adequate (Calculated Using Formula 5) column filled with smaller particles. The RSLC method Theoretical Plates N is predicted best if the selectivity of the stationary phase Particle size 4.5 µm 3 µm 2.2 µm is maintained. Therefore, a column from the same Column length: 250 mm 18518 27778 37879 manufacturer and with nominally identical surface 150 mm 11111 16667 22727 modification is favoured for an exact method transfer. If this is not possible, a column with the same nominal 100 mm 7407 11111 15152 stationary phase is the best choice. The separation is made 75 mm 5555 8333 11364 faster by using shorter columns, but the column should 50 mm 3703 5556 7576 still offer sufficient column efficiency to allow at least a baseline separation of analytes. Table 1 gives an overview If the resolution of the original separation is higher of the theoretical plates expected by different column than required, columns can be shortened. Keeping the length and particle diameter size combinations using column length constant while using smaller particles Dionex Acclaim column particle sizes. Note that column improves the resolution. Reducing the column diameter manufacturers typically fill columns designated 5 µm with does not shorten the analysis time but decreases mobile particle sizes 4–5 µm. Dionex Acclaim 5 µm columns are phase consumption and sample volume. Taking into actually filled with 4.5 µm particles. This is reflected in account an elevated temperature, smaller column inner the table. diameters reduce the risk of thermal mismatch.

119 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator System Requirements Table 2. Mixer Kits Available for UltiMate 3000 Smaller particles generate higher backpressure. The RSLC System to Adapt GDV of Pump linear velocity of the mobile phase has to be increased Mixer Kit GDV pump while decreasing the particle size to work within the Van Mixer kit 6040.5000 35 µL Deemter optimum. The UltiMate 3000 RSLC system Static mixer kit 6040.5100 100 µL perfectly supports this approach with its high maximum Static mixer kit 6040.5150 200 µL operation pressure of 800 bar (11,600 psi). This maximum pressure is constant over the entire flow rate range of up to 5 mL/min, providing additional potential to speed up bypass mode. The GDV of a standard sample loop of the applications even further by increasing the flow rate. RSLC autosampler is 150 µL, the micro injection loop has a 50 µL GDV. Besides the gradient delay volume, the extra column volume is an important parameter for fast LC methods. Gradient Delay Volume The extra column volume is the volume in the system Detector through which the sample passes and hence contributes Autosampler Column to the band broadening of the analyte peak (Figure 3). Pump Extra Extra Column Column The extra column volume of an optimized LC system Volume Volume should be below 1/ th of the peak volume. Therefore 25717 10 the length and inner diameter of the tubing connections Figure 3. Gradient delay volume and extra column volume of an from injector to column and column to detector should HPLC system. Both play an important role in method speed-up. be as small as possible. Special care has to be taken while installing the fittings to avoid dead volumes. In addition, the volume of the flow cell has to be adapted to the peak For fast gradient methods, the gradient delay volume volumes eluting from the RSLC column. If possible, the 1 (GDV) plays a crucial role. The GDV is defined as the flow cell detection volume should not exceed /10th of the volume between the first point of mixing and the head of peak volume. the column. The GDV becomes increasingly important with fast, steep gradients and low flow rate applications as Detector Settings it affects the time taken for the gradient to reach the head When transferring a conventional method to an RSLC of the column. The larger the GDV, the longer the initial method, the detector settings have a significant impact isocratic hold at the beginning of the separation. Typically, on the detector performance. The data collection rate and this leads to later peak elution times than calculated. Early time constant have to be adapted to the narrower peak eluting peaks are affected most. In addition, the GDV shapes. In general, each peak should be defined by at least increases the time needed for the equilibration time at the 30 data points. The data collection rate and time constant end of a sample and therefore increases the total cycle settings are typically interrelated to optimize the amount time. A general rule is to keep the gradient steepness of data points per peak and reduce short-term noise while and the ratio of GDV to column volume constant when still maintaining peak height, symmetry, and resolution. transferring a standard method into a fast LC method. The Chromeleon® Chromatography Management This will maintain the selectivity of the original method.3 Software has a wizard to automatically calculate the The GDV can be adjusted to the column volume by best settings, based on the input of the minimum peak installing appropriate mixer kits to the RSLC pump (see width at half height of the chromatogram. This width Table 2), which contributes most to the GDV. Typically, is best determined by running the application once at 100 µL or 200 µL mixers are good starting points when maximum data rate and shortest time constant. The operating a small volume column in an RSLC system. obtained peak width may then be entered into the wizard Another option is to switch the sample loop of the for optimization of the detection settings. Refer to the split-loop autosampler out of the flow path. The GDV is detector operation manual for further details. then reduced by the sample loop volume in the so-called

120 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Method Speed-Up Using the Calculator Peaks: 1. Acesulfame K 5. Benzoate Separation Example 2. Saccharin 6. Sorbate 3. Caffeine 7. Benzaldehyde Separation was performed on an UltiMate 3000 4. Aspartame RSLC system consisting of a HPG-3200RS Binary Rapid 4.6 × 150 mm, 4.5 µm 1800 A 1.5 mL/min 5 Separation Pump, a WPS-3000RS Rapid Separation Well 6 2 3 Plate Sampler with analytical sample loop (100 µL), a 1 mAU TCC-3000RS Rapid Separation Thermostatted Column 4 7 Compartment with precolumn heater (2 µL), and a VWD-3400RS Variable Wavelength Detector with semi- –200 1000 micro flow cell (2.5 µL). Chromeleon Chromatography B 5 Management Software (version 6.80, SR5) was used 6 3 2.1 × 50 mm, 2.2 µm for both controlling the instrument and reporting the mAU 1 7 0.639 mL/min 2 4 data. The modules were connected with stainless steel 1 micro capillaries, 0.01” ID, /16" OD when applying the –100 conventional LC method, 0.007” and 0.005” ID, 1/ " OD 16 1000 C when applying the RSLC methods. A standard mixture 5 6 2.1 × 50 mm, 2.2 µm of seven common soft drink additives was separated by 3 1.599 mL/min mAU 12 7 gradient elution at 45 °C on two different columns: 4 • Conventional HPLC Column: Acclaim 120, C18, –100 5 µm, 4.6 × 150 mm column, (P/N 059148) 0 2 4 6 8 10 Minutes • Rapid Separation Column: Acclaim RSLC 120, C18, 25718 2.2 µm, 2.1 × 50 mm column (P/N 068981). Figure 4. Method transfer with the Method Speed-Up Calcula- tor from A) a conventional LC separation on an Acclaim 5 µm The UV absorbance wavelength at 210 nm was particle column, to B) and C) RSLC separations on an Acclaim recorded at 5 Hz using the 4.6 × 150 mm column and at 2.2 µm particle column. 25 Hz and 50 Hz using the 2.1 × 50 mm column. Further method details such as flow rate, injection volume, and gradient table of conventional and RSLC methods are resolution is R(5,6)=3.48. This resolution is sufficiently high described in the following section. The parameters for the to select a fast LC column with fewer theoretical plates for method transfer were calculated with the Dionex Method the speed up. Therefore, a 2.1 × 50 mm, 2.2 µm column Speed-Up Calculator (version 1.14i). with 7579 plates was selected. The conventional separation of seven soft drink The first values to be entered into the yellow field additives is shown in Figure 4A. With the Method Speed- of the Method Speed-Up Calculator are the current Up Calculator, the method was transferred successfully to column dimension, planned column dimension, and RSLC methods (Figure 4B and C) at two different flow the resolution of the critical pair. To obtain the most rates. The easy method transfer with this universal tool is accurate method transfer, use the particle sizes listed in described below. the manufacturer's column specifications sheet instead of the nominal size, which may be different. Dionex Column Selection for Appropriate Resolution Acclaim columns with a nominal particle size of 5 µm are The column for method speed-up must provide actually filled with 4.5 µm particles, and this value should sufficient efficiency to resolve the most critical pairs. be used to achieve a precise method transfer calculation. In this example, separating peaks 5 and 6 is most This has a positive impact on the performance and challenging. A first selection of the planned column pressure predictions for the planned column. Based on dimensions can be made by considering the theoretical the assumption of unchanged stationary phase chemistry, plates according to Table 1. The 4.6 × 150 mm, 5 µm the calculator then predicts the resolution provided by the column is actually filled with 4.5 µm particles. Therefore, new method (Figure 5). it provides 11,111 theoretical plates. On this column, the

121 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Figure 5. Column selection considering the resolution of the critical pair.

Figure 6. The flow rate, injection volume and backpressure of the current method are scaled to the new column dimension.

In the example in Figure 5, the predicted resolution planned method. In addition, the new flow rate is scaled between benzoate and sorbate is 2.87. With a resolution of to the change of column cross section if the column R ≥1.5, the message “Baseline resolution achieved” pops inner diameter changed. This keeps the linear velocity up. This indicates that a successful method transfer with of the mobile phase constant. A boost factor (BF) can be enough resolution is possible with the planned column. entered to multiply the flow rate for a further decrease If R is smaller than 1.5, the red warning “Baseline is not in separation time. If the calculated resolution with resolved” appears. Note that the resolution calculation is BF=1 predicts sufficient separation, the method can be performed only if the boost factor BF is 1, otherwise it is accelerated by increasing the boost factor and therefore disabled. The function of the boost factor is described in increasing the flow rate. Figure 1 shows that applying the Adjust Flow Rate section. linear velocities beyond the optimum is no problem with smaller particle phases, as they do not significantly loose Instrument Settings plates in this region. Note that the resolution calculation The next section of the Method Speed-Up Calculator of the Method Speed-Up Calculator is disabled for BF≠1. considers basic instrument settings. These are flow rate, For the separation at hand, the flow rate is scaled injection volume, and system backpressure of the current from 1.5 mL/min to 0.639 mL/min when changing method (Figure 6). In addition to these values, the detector from an Acclaim 4.6 × 150 mm, 4.5 µm column to a settings have to be considered as described in the earlier 2.1 × 50 mm, 2.2 µm column (see Figure 6), adapting the section “Detector Settings”. Furthermore, the throughput linear velocity to the column dimensions and the particle gain with the new method can be calculated if the number size. The predicted resolution between peak 5 and 6 for of samples to be run is entered. the planned column is R=2.87. The actual resolution achieved is R=2.91, almost as calculated (chromatogram Adjust Flow Rate B in Figure 4). As explained by Van Deemter theory, smaller particle A Boost Factor of 2.5 was entered for further phases need higher linear velocities to provide optimal acceleration of the method (Figure 7). The method was separation efficiency. Consequently, the Dionex Method then performed with a flow rate of 1.599 mL/min, and Speed-Up Calculator automatically optimizes the linear resolution of the critical pair was still sufficient at R=2.56 velocity by the ratio of particle sizes of the current and (see zoom in chromatogram C in Figure 4).

122 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Figure 7. The new flow rate is further accelerated by applying the Boost Factor of 2.5.

Scale Injection Volume of mobile phase is considered constant during method The injection volume has to be adapted to the new transfer. The calculated pressure is only an approximation column dimension to achieve similar peak heights by and does not take into account nominal and actual particle equivalent mass loading. Therefore the injection plug size distribution depending on column manufacturer. has to be scaled to the change of column cross section. If the predicted maximum pressure is above 800 bar In addition, shorter columns with smaller particles cause (11,600 psi) the warning “Exceeds pressure limit RSLC” a reduced zone dilution. Consequently, sharper peaks is shown, indicating the upper pressure limit of the compared to longer columns are expected. The new UltiMate 3000 RSLC system. However, in the case the

injection volume Vinj,new is then calculated by Formula 8, method is transferred to a third party system, its pressure taking a changed cross section and reduced band specification has to be considered. broadening by changed particle diameter into account. In the example of the soft drink analysis, the actual pressure increases from 92 bar to 182 bar with BF=1 on

2 the 2.1 × 50 mm column, and to 460 bar for the RSLC d ⋅ dL Formula 8: ,newc ,newpnew method with BF=2.5. The pressures predicted by the ,newinj = VV ,oldinj ⋅⋅ ( d ( ⋅ dL ,oldc ,oldpold Method Speed-Up Calculator are 262 bar and 656 bar, respectively. The pressure calculation takes into account Generally, it is recommended that a smaller flow cell the change of the size of the column packing material. In a be used with the RSLC method to minimize the extra speed up situation, the pressure is also influenced by other column volume. Also, the difference in path length of factors such as particle size distribution, system fluidics different flow cell sizes has to be taken into account while pressure, change of flow cell, etc. When multiplication scaling the injection volume. In the example of the soft factors such as the boost factor are used, the difference drink analysis, the injection volume is scaled from 25 µL between calculated and real pressure is pronounced. to 2.1 µL when replacing the Acclaim 4.6 × 150 mm, The pressure calculation is meant to give an orientation, 4.5 µm column with a 2.1 × 50 mm, 2.2 µm column what flow rates might be feasible on the planned column. (see Figure 6). However, it should be confirmed by applying the flow on the column. Predicted Backpressure Speeding-up the current method by decreasing Adapt Gradient Table particle size and column diameter and increasing flow rate The gradient profile has to be adapted to the changed means elevating the maximum generated backpressure. column dimensions and flow rate following the gradient- The pressure drop across a column can be approximated volume principle. The gradient steps of the current by the Kozeny-Carman formula.4 The pressure drop of method are entered into the yellow fields of the gradient the new method is predicted by the calculator considering table. The calculator then scales the gradient step changes in column cross section, flow rate, and particle intervals appropriately and creates the gradient table size and is multiplied by the boost factor. The viscosity of the new method.

123 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Figure 8. The gradient table of the current method (A) is adapted to the boosted method (B) according to the gradient-vol- ume principle.

Figure 9. The absolute values for analysis time, eluent usage, and sample usage of the current (purple) and planned (green) method are calculated by the Method Speed-Up Calculator. The savings of eluent, sample, and time due to the method transfer are highlighted.

The adapted gradient table for the soft drink analysis Consumption and Savings while using a boost factor BF=1 is shown in Figure 8. Why speed-up methods? To separate analyte peaks According to the gradient-volume principle, the total run faster and at the same time reduce the mobile phase and time is reduced from 29.0 min to 4.95 min by taking into sample volume consumption. Those three advantages of account the changed column volume from a 4.6 × 150 a method speed-up are indicated in the Method Speed- mm, 5 µm (4.5 µm particles entered) to a 2.1 × 50 mm, Up Calculator sheet right below the gradient table. The 2.2 µm column and the flow rate reduction from absolute values for the time, eluent, and sample usage are 1.5 mL/min to 0.639 mL/min. The separation time was calculated taking the numbers of samples entered into the further reduced to 1.89 min by using boost factor BF=2.5. current instrument settings section of the calculation sheet Gradient time steps were adapted accordingly. The into account (see Figure 6). comparison of the peak elution order displayed in Regarding the soft drink analysis example, Figure 4 shows that the separation performance of the geometrical scaling of the method from the conventional gradient was maintained during method transfer. column to the RSLC method means saving 93% of eluent and 92% of sample. The sample throughput increases 6.1-fold using BF=1. The higher flow rate at BF=2.5 results in a 15.3-fold increased throughput compared to the conventional LC method (Figure 9).

124 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Conclusion References Fast method development or increased sample 1. Halász, I.; Endele, R.; Asshauer, J. J. Chromatogr., A throughput are major challenges of most analytical 1975, 112, 37–60. laboratories. A systematic method speed-up is 2. Snyder, L.R.; Dolan, J.W.; Grant, J.R. accomplished by reducing the particle size, shortening J. Chromatogr., A 1979, 165, 3–30. the column length, and increasing the linear velocity 3. Schellinger, A.P.; Carr, P.W. J. Chromatogr., A 2005, of the mobile phase. The Dionex Method Speed-Up 1077, 110-119. Calculator automatically applies these rules and scales 4. Bear, J. Dynamics of Fluids in Porous Media; Dover: the conventional LC parameters to the conditions of Mineola, NY, 1988. the RSLC method. The interactive electronic tool is universally applicable. New instrument settings are predicted and gradient tables are adapted for optimum performance for the new method. The benefit of the method transfer is summarized by the integrated calculation of savings in time, eluent and sample. In addition, users can benefit from getting results earlier and thereby reducing the time to market. The Dionex Method Speed-Up Calculator is part of Dionex’s total RSLC solution, which further consists of the industry leading UltiMate 3000 RSLC system, powerful Chromeleon Chromatography Management Software, and high- efficiency Acclaim RSLC columns.

125 Easy Method Transfer from HPLC to RSLC with the Dionex Method Speed-Up Calculator Alcon OPTI-FREE RepleniSH is a registered trademark of Alcon Inc., Hünenberg, Switzerland. Bacto is a registered trademark of Difco Laboratories GmbH. Bausch & Lomb Gentle Sensitive Eyes is a registered trademark of Bausch & Lomb Inc., Rochester, NY. Humatin is a registered trademark of Monarch Pharmaceuticals. Hyflo Super Cel is a registered trademark of Manville Corp. Milli-Q and Millex are registered trademarks of Millipore Corporation. Neosporin is a registered trademark of Pfizer Consumer Healthcare. Maxipime is a registered trademark of Bristol-Myers Squibb Co. NeoDecadron is a registered trademark of Merck & Company, Incorporated. PediOtic is a registered trademark of Monarch Pharmaceuticals, Incorporated. PEEK is a registered trademark of Victrex PLC. Tween is a registered trademark of Atlas Chemical Co. USP is a registered trademark of the United States Pharmaceopeia.

All other trademarks are the property of Thermo Fisher Scientific Inc. and its subsidiaries.

©2012 Thermo Fisher Scientific, Inc. All rights reserved. This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

Dionex Products: 1228 Titan Way, PO Box 3603, Sunnyvale, CA 94088-3603, (408) 737-0700 North America: U.S./Canada (847) 295-7500 South America: Brazil (55) 11 3731 5140 Europe: Austria (43) 616 51 25, Benelux (31) 20 683 9768 (32) 3 353 4294 Denmark (45) 36 36 90 90, France (33) 1 39 30 01 10, Germany (49) 61125 991 0 Ireland (353) 644 0064, Italy (39) 02 51 62 1267, Sweden (46) 8 473 3380, Switzerland (41) 62 205 9966, United Kingdom (44) 1276 691722 Asia Pacific: Australia (61) 2 9420 5233, China (852) 2428 3282, India (91) 22 2764 2735, Japan (81) 6885 1213, Korea (82) 2 2653 2580, Singapore (65) 6289 1190, Taiwan (886) 2 875 6655 www.thermoscientific.com/dionex LPN 3035 04/12