2014 Technical Posters

CoSMoS 2014 posters


Poster 1

Retention ‘Projection’ Enables Reliable Use of Shared HPLC Metabolite Retention Data Across Labs and Methods

Daniel Abate-Pella

University of Minnesota, MN

Use of retention times of standards to measure and account for differences between HPLC instruments in seven different labs

Our limited ability to identify molecules is a major bottleneck in metabolomics. Supplementing MS information with LC retention data would allow many more metabolites to be identified, but this data is very difficult to utilize reliably because it is influenced by many experimental factors that differ from lab to lab – some differences are intentional and some are not. A new approach we call ‘retention projection’ accounts for both types of differences. The methodology first measures the actual gradient profile produced by an HPLC instrument using the retention times of a set of benzamide standards, and then uses that information to calculate (or ‘project’) the retention times of other metabolites of interest. In previous work, we ran 20 chemically diverse molecules in different gradients, flow rates, and on different LC instruments, and found that the accuracy of projected retention times were on the level of that expected from theory. In this work, we set out to test the practical accuracy of the system by conducting an inter-laboratory study with seven other labs. We asked each lab to carry out ten gradients: a set of four linear ones and six that they chose themselves. The root mean square projection errors for the first four gradients were consistent with our own results: ±0.88 s for a 5 min gradient, ±3.06 s for 30 min, and ±1.51 and ±1.65 s for 10 min ones at different flow rates. This is more accurate than indices errors of ±1.54, ±4.99, ±2.23 and ±2.47 s. The accuracy of the methodology remained similarly high under all of the conditions tested in the six unique gradients each lab performed; the highest errors were observed in methods with isocratic steps (which is expected from theory) or when one or more of the benzamide standards were not detected. We have started building a larger metabolite database, available online at www.retentionprediction.org, to use this approach.


Poster 2

Development of an Instrument and Method Independent HPLC Retention Database for Enhanced Metabolite Identification

Josh Hewitt

University of Minnesota, MN

Development of an instrument and method independent HPLC retention database for enhanced metabolite identification

We are creating an open, freely accessible repository of isocratic retention data for a wide-ranging selection of biologically relevant metabolites. This data, when combined with freely available, open-source software, allows for the prediction of gradient retention times with uniquely high accuracy. Moreover, the methodology is applicable with virtually any HPLC instrument and under a wide range of experimental conditions (i.e. different gradients, flow rates, and column dimensions). Briefly, a set of standards are run in a fast gradient to measure the actual gradient profile produced by an instrument. This information is then loaded into the open-source software at which point the isocratic retention information is used to predict the retention time of a given compound. Using this approach, retention times accurate to within 0.23% to 0.87% of the gradient time can be predicted. When coupled with mass spectral data, retention times predicted with this degree of accuracy allow for greatly enhance metabolite identification. For example, in a simulation using 7307 compounds from the KEGG database a single quadrupole MS is able to identify (99% confidence) approximately 1% of metabolites using MS data alone. When predicted retention times accurate to 0.25% are included the portion of compounds increases to 44%. At the time of this writing, we have compiled a library of more than 500 metabolite standards. Scouting gradient runs have been completed on close to 200 of these samples, which allows us to establish groupings of compounds with similar retention times and thereby reduce the number of necessary measurements. We have collected retention data for 70 compounds and expect to have more than 500 compounds in the database by August. The isocratic data is freely available at www.retentionprediction.org/hplc/database/index.php as online content or a downloadable .CSV file.


Poster 3

Development and Validation of a Method for the Determination of Trace 2-Hydroxyethylhydrazine in a Drug Substance by LC-MS

Cale Bibb

Array BioPharma Inc., CO

A reversed-phase HPLC method to determine trace amounts of 2-hydroxyethylhydrazine using an internal standard and mass spectrometric detection

2-Hydroxyethylhydrazine is a common building block for the synthesis of drug substances. However, it is a known genotoxic compound and must be controlled in drug substances to very low levels. A method to determine trace amounts of 2-hydroxyethylhydrazine has been developed and validated. The method uses an internal standard of d4-2-hydroxyethylhydrazine and analysis by reversed-phase HPLC with mass spectrometric detection. The method gives excellent sensitivity with a limit of quantitation (LOQ) of 0.5 ppm when the drug substance samples are prepared at 100 mg/mL. The spike recoveries of 2-hydroxyethylhydrazine at a level of 1.5 ppm were between 92 and 116%. A linear range of concentrations from 0.5 to 5 ppm has been demonstrated, with R2 > 0.99. Additionally, the method has been demonstrated for drug substance samples prepared at 25 mg/mL. The two level method approach allows for the determination of a wide range of 2-hydroxyethylhydrazine concentrations using a single validated method. Spike recoveries at a level of 20 ppm were between 97 and 120%.


Poster 4

A Study of the Degradation Mechanism of a 1,2,4-Oxadiazole Moiety to a Nitrile Group

Yande Huang

Bristol-Myers Squibb, NJ

Using benzylamine to probe the degradation mechanism of the 1,2,4-oxadiazole ring to the nitrile group

BMS-708163 is a Gamma-secretase inhibitor drug candidate that contains a 1,2,4-oxadiazole moiety. During the BMS-708163 stability study, it was found that the 1,2,4-oxadiazole moiety degraded under stress conditions to a nitrile group either in solution or solid state. The degradation of the 1,2,4-oxadiazole moiety was relatively slow at a pH range of 3 to 5, but increased significantly at lower and higher pH ranges. 15N isotope labeling on the N2 of the 1,2,4-oxadiazole ring was used to investigate the degradation mechanism by LC-MS and indicated that the nitrile group was derived solely from the N2 of the 1,2,4-oxadiazole ring. However, it was still not clear how the degradation initiated. Therefore, benzylamine was used as an organic base to catalyze the degradation of the 1,2,4-oxadiazole ring to the nitrile group and trap the remaining portion of the 1,2,4-oxadiazole ring. The forced degradation reaction was monitored using 1H NMR. It was unexpected to find that the methine proton of the 1,2,4-oxadiazole ring underwent H/D exchange in either CD3OD or CD3CN/D2O. This finding was then used to probe the protonation of the 1,2,4-oxadiazole ring at the N4 position which provided solid evidence to support the proposed degradation mechanism. Details of the investigation and rationale of the degradation mechanism will be presented.


Poster 5

Tracking isobaric metabolites across changing chromatographic methods and modes using ion mobility

Mark Wrona

Waters Corporation, MA

Tracking isobaric peaks which change elution order across chromatographic modes using ion mobility

Tracking the fate of multiple (isobaric) metabolites under varying chromatographic (RT and elution order changing) conditions is a challenge during drug development. Changes in analytical methodology can lead to different orders of elution, which becomes especially problematic for tracking isobaric metabolites. In this abstract we describe the use of m/z-ion mobility based data pairs to help correlate metabolites across varying UHPLC and SFC chromatographic modes with completely different elution orders.

Model compounds (including Carbamazepine) were incubated in human liver microsomes at two time points (0 and 60 minutes). Incubations were quenched at specified time points, protein precipitated, and centrifuged. Supernatant was analyzed using several techniques, including different separation modes (UPLC and SFC) and with varying gradients (UPLC based). In all cases, MS information was recorded by coupling the chromatographic systems (Acquity UPLC or Acquity UPC2) to a Synapt G2-Si mass spectrometer capable of recording HDMSE (low and high energy fragment time aligned data recorded with ion mobility information). Drift times for all LCMS ions were automatically recorded during the acquisition.

For carbamazepine, two major +O metabolites were characterized, M1 and M2. M1 was observed with an average drift time of 1.906 +/- 0.078 (0.4%) ms (range 1.896-1.919 ms). For M2, an average of 1.874 +/- 0.012 ms (0.64%) ms (range 1.851-1.884) was observed. These measurements were made across three different UPLC gradients and one UPC2 gradient. The precision of measurements was high enough to different the two isobaric species regardless of their elution order.

The ability to have ion mobility as an additional measurement is a useful physical property to enhance and correlate the information derived from various HRMS analyses.


Poster 6

Automated Structure Elucidation Of Tandem Mass Spectra For Unknown Metabolites And Other Small Molecules Using The Masspec Algorithm

Marshall Siegel

MS Mass Spec Consultants, NJ

A unique and powerful algorithm, MASSPEC, was developed to automatically interpret/correlate mass spectral data with chemical structures

The MASSPEC algorithm was designed and implemented to elucidate/correlate chemical structures with observed tandem mass spectra acquired in exact mass or nominal mass modes produced under any ionization technique. The program is ideally suited to analyze

known and unknown small molecules. For example, the structures of metabolites from metabolomics and pharmaceutical studies can be readily elucidated since the product structures are based upon classes of known starting natural products and pharmaceutical structures, each producing a variety of possible metabolic modifications whose site in the molecules can be readily identified by the algorithm.


The MASSPEC algorithm is written in Visual Basic and is based upon advanced graph theory and combinatorial methods. I/O features include a Data Input Module and an Output Graphics Display Module. The metabolomics and metabolite data were acquired in the ESI/MSMS mode of the [M+H]1+ parent ion in the exact mass mode (rp = 18,000,  = 5 ppm). The proposed structures for the fragment ions were obtained by using the MASSPEC algorithm to automatically compute the highest scoring fragment ion substructures for the observed ions.

Preliminary Data

The underlying principle of the MASSPEC algorithm is that a proposed chemical structure is described by connected sets of atoms that do not fragment further, where each set is referred to as a superatom. The chemical structure is therefore a connected set of superatoms, where the connections are viewed as the chemical bonds linking the superatoms. A mass spectrum is generated from ionized substructures of the original parent molecule, whereby the connectivities of the superatoms in the original molecule remain intact, are broken or are modified by predicted rearrangements. In addition, substructures, referred to as floating superatoms, are incorporated into the calculation. These floating superatoms can be located anywhere within the molecule but are localized by determining their presence in each of the fragment ions. The floating superatoms can serve as additions or losses of substructures to a superatom, thereby behaving as metabolic changes to the parent structure. A scoring system, based upon the number of bonds cleaved and the mass accuracy of the observed vs. predicted masses are used to predict the best possible chemical structures for the fragment ions from a number of possibilities. Applications and strategies for use of the algorithm will be illustrated for two metabolomics problems of unknown metabolites related to norcardamine and tetracycline and for three primary metabolites from three different pharmaceuticals and a variety of small molecules. The MASSPEC algorithm exhaustively computes all the possible substructures, limited only by the input constraints, and correlates them with the observed ions, a task which is virtually impossible to achieve by hand calculations.


Poster 7

Elucidation of Polysaccharide Structures, Containing Charged Polysulfated/Polycarboxylated Moieties, Using Exact-Mass ESI Negative Ionization MSMS Data and the MASSPEC Algorithm

Marshall Siegel

MS Mass Spec Consultants, NJ

Complex polysaccharide structures are fully elucidated using MSMS data and the MASSPEC structure elucidation algorithm

Arixtra and dP7 are two members of a class of highly charged and highly sulfated linear polysaccharides. These compounds produce very complex spectra in the ESI negative ionization MSMS mode that is challenging to correlate with the structures. The MASSPEC algorithm, derived from Graph Theory and combinatorics, was used to correlate the fragment ions observed in the ESI/MSMS spectra with corresponding substructures of the proposed structures.


Arixtra (C31H53N3O49S8, M = 1506.95134 Da) is a pentasaccharide containing 8 sulfo and 2 carboxyl groups.dP7 (C44H71N3O57S7, M = 1777.07943 Da) is a septasaccharide containing 7 sulfo and 3 carboxyl groups. Each compound has a total of 10 acidic sites. All ionizable protons were replaced with sodium during the electrospray process using a carrier solvent containing NaOH. Exact mass data were acquired (D = 3 ppm ) with an Orbitrap mass spectrometer in the ESI/MSMS negative ionization mode for the [M-10H+7Na]3- parent ions at m/z 552.601 for Arixtra and m/z 642.6423 for dP7. The proposed structures for the fragment ions were obtained by using the MASSPEC algorithm that automatically computed the highest scoring fragment ion substructures for the observed ions.

Preliminary Data

The Arixtra and dP7 raw data were pre-processed for analysis by the MASSPEC algorithm by removal of chemical and electronic noise, converting all multiply charged negative ions to singly charged negative ions using Na+ as the counter ion and removal of all isotopic ions. The Arixtra and dP7 structures were each re-expressed as sets of bonded and floating superatom substructures. Expected fragmentation rearrangements were expressed in the organization of the bonded superatom structures and through the use of floating superatoms. The MASSPEC algorithm automatically computed the masses of all possible substructures with up to four bonds removed and compares the predicted masses to the observed masses in the MSMS spectrum. The MASSPEC scoring algorithm incorporates MS fragmentation processes and degree of fit of the predicted and observed masses. The most probable fragment ion assignments were the highest scoring predicted substructures for the fragment ion masses. Additional filtering of the solutions were performed to retain only those solutions with a 1- charge and no excised superatoms, i.e., sugar rings containing two or more bonded superatom ring members. Using the MASSPEC algorithm, in this fashion, reasonable assignments were made for all the observed fragment ions, including those generated from secondary fragmentation processes. This structure elucidation work is virtually impossible to be successfully achieved by hand calculations since only with computer based algorithms are the calculations exhaustive. This methodology is adaptable for identifying the sites of modifications in parent polysaccharides and to unknown polysaccharide structures.


Poster 8

Square Wave Stripping Voltammetric Determination Of Histamine In The Presence Cu(Ii) Ions And Application To Fish Samples

Ummihan Taskoparan Yilmaz

Gazi University, Ankara, Turkey

A new method has been developed for the determination of histamine using electrochemical method

A square wave stripping voltammetric (SWSV) method has been developed for the indirect determination of histamine. The complex of copper(II) and histamine gives a well defined cathodic stripping peak current at −420 mV, which has been used for the determination of histamine with accumulation time of 10 s at 0.0 V versus Ag/AgCl. The optimum conditions of pH, accumulation potential and accumulation time were studied. Two linear calibration graphs were obtained with slopes of 0.078 (µM/µA) and 0.014 (µM/µA), respectively. The detection limits were found to be 3×10-7 M and 1×10-5 M for histamine (S/N = 3), respectively The effect of common excipients and metal ions on the peak height of Cu-histamine complex peak was studied. The method was successfully, applied to the determination of histamine in canned anchovy, frozen tinca tinca and cyprinus carpio fish samples.


Poster 9

Infrared Analysis: The Silver Hammer in Maslow’s Toolbox

Douglas Minick

GlaxoSmithKline, NC

This poster describes the capability of ab initio IR calculations to elucidate structures largely by pattern recognition, mitigating data analysis by an experienced vibrational spectroscopist.

Results will be presented for two ab initio IR analyses recently conducted in my laboratory, the regio chemistry analysis of a set of dimethyl-amino-1H-pyrrazole-1-carboxylates, and confirmation of the relative stereochemistry expected for a commercially available set of cis/trans 3-hydroxycyclobutanecarboxylates. These examples illustrate two structure problems commonly encountered in small molecule synthesis, both of which can be problematic for NMR: elucidating the regio chemistry of heterocycles and the relative stereochemistry of disubstituted saturated rings with a plane of symmetry. Basic methods will be described for synthesizing the conformationally averaged IR spectra used in these studies, including details of the conformational search parameters (molecular mechanics/MMFF94x force field/Born solvation/search modes), quantum mechanical methods used to calculate harmonic frequencies and vibrational intensities, (e.g., B3LYP functional with various basis sets for increasing accuracy). Calculated spectra will then by compared with solid state ATR-IR spectra (cast films) and solution phase IR absorption spectra to identify the best set of computational and experimental conditions giving the best match. The level of confidence in the assignments will then be quantified using CompareVOATM, a commercial algorithm developed for ab initio vibrational CD analysis. Results from both studies will show that all assignments are conclusive, demonstrating the high level of reliability attainable using vibrational spectra calculated by DFT methods.


Poster 10

Investigation of the Mechanism of Interaction between Glatiramer Acetate and Quinoline-moiety containing Compound

Helena Cooperman

Teva Pharm Industries, Ltd, Israel

Investigation of the complex mixtures by modern analytical techniques

Glatiramer acetate (GA) is the active ingredient of COPAXONE (Teva Pharmaceutical Industries Ltd), a synthetic polymeric drug, widely used for the treatment of multiple sclerosis.

Copaxone may be characterized as a complex colloidal mixture composed of the immunogenic polypeptides. An elucidation of the principle of self-assembly processing of the molecules in the GA polypeptides mixture is important for more complete understanding of its mechanism of action.

One of the analytical tools for investigation of the secondary and tertiary structure in the polypeptidic molecules is the tracing of conformational changes in the molecules caused by interaction with small hydrophobic molecules, like fluorescent dyes and Quinoline-moiety containing compounds. Quinoline derivatives are known as the potent protein-modulators and utilized also as the gelation agent in ethanol solutions.

In the current study, a Quinoline-moiety containing chemical compound (QMCC) was found to be a very efficient agent for promotion of aggregation in GA solutions.

The interactions between polymer and QMCC were studied by a variety of analytical techniques: SEC HPLC, Mass-spectroscopy on Instrument Triple ToF 5600 MS/MS AB SCIEX, CD and Cryo-TEM-microscope techniques. The fractionation of the GA-QMCC gel-like solutions was performed by ultracentrifugation.

It was shown, that the kinetic of the gelation in Glatiramer Acetate solution was strongly correlated with elevation of temperature, prolongation of incubation time and mainly, on the concentration of QMCC in the tested mixtures.

It was demonstrated, that reassembling of the GA polypeptides molecules is accompanied by appearance of hydrophobic nuclei and formation of the fibrous structures which entangle to form a three-dimensional network of gels.

The implementation of Cryo-TEM-microscope technique revealed, that the fiber density in the precipitated fraction of GA-QMCC solution is significantly higher than that in the liquid phase. Implementation of the other analytical method showed, that fibers in the hydrogel are formed via assembly through non-covalent forces.

Amino acid analysis of both fractions established, that the precipitated fraction of GA /QMCC solution has a different amino-acid composition compared to the supernatant. The concentration of Tyr-residues in the precipitated fraction is about 20% higher than in the supernatant.

This finding supports our assumption on the hydrophobic nature of the interaction between Glatiramer Acetate and Quionoline-moiety containing compound.


Poster 11

Automated selection of optimal purification conditions from LC/MS and SFC/MS screening methods

Joseph Simpkins

Virscidian, Cary, NC

Screening methods are a high throughput common approach for accelerating method development for chromatographic separations. Using this approach, several method conditions can be screened in serial or parallel to reduce or eliminate timely method development where chromatographic peaks. However, as the numbers of method conditions are increased, inevitably the amount of time to review and select the optimal condition increases. Utilizing the Analytical Studio Express Server and tunable functions or expressions with the Analytical Studio suit of software, a flexible and dynamic method selection package was implanted to analyze and review both SFC and HPLC screening conditions.


Supercritical Fluid (SFC) and High pressure liquid (HPLC) chromatographic systems were interfaced to single quadrupole mass spectrometric detection.  The SFC and HPLC systems consisted of Agilent 1100 LC-MSD’s which include quaternary or binary pumping components, autosampler, DAD detectors and G1946D MSD Systems.  The SFC system also included an Aurora Fusion A5 unit to condition carbon dioxide for use in these experiments. The automated software utilized Analytical Studio Express and Analytical Studio software from Virscidian with internal macro functions and customized visualizations.
Preliminary data

The software was validated against 100 samples utilizing 6 SFC and 4 HPLC screening methods.  The results of the screen selected the best method [95%] of the time for SFC and [95%] of the time for HPLC.  The subsequent methods were purified and good success.  Additionally, the increase in throughput for lab personnel resulted in an overall drop in turnaround time by 15%.  Figure1 shows the visualization of the Analytical studio output for a typical sample.


Poster 12

Drug Reactivity with formaldehyde in an early development immediate-release (IR) tablet formulation

Todd Zelesky

Pfizer, CT

Multidisciplinary approach key to understanding product distribution of MS/NMR characterized regioisomeric drug-formaldehyde degradation products

Reactive excipient impurities can compromise the chemical stability of a drug substance, impact product performance and compromise quality. Formaldehyde is a highly reactive excipient impurity ubiquitous to many excipients used in the manufacture of formulated drug products. We report the reactivity of a drug substance with formaldehyde, in an early development immediate-release (IR) tablet formulation. Multiple regioisomeric degradation products are discussed in this interesting case study. A drug substance-formaldehyde enrichment strategy was employed to support SFC isolation and MS/NMR characterization of targeted degradation products.


Poster 13

Computational ranking of plant compounds based on antimicrobial class against methicillin resistant Staphylococcus aureus

Caroline Rempe

University of Tennessee, Knoxville, TN

A list of unidentifiable plant compounds with known mass spectral tags and the potential to inhibit MRSA was generated.

Antibiotic resistance is a significant problem for human and animal health with the development and production of novel antimicrobials lagging behind the evolution of multi-drug resistant bacteria. Bioactive plant compounds are a potential source for novel antimicrobial formulations, so we examined the composition of antimicrobial active fractions of an aqueous botanical extract (not named for intellectual property purposes) known to have antimicrobial activity against methicillin resistant Staphylococcus aureus (MRSA). Solvent based fractionation was used to divide the aqueous botanical extract into antimicrobial active and inactive components. Trimethylsilyl derivatives of fractions were then analyzed by gas chromatography – mass spectrometry (GC-MS) with electron impact ionization to generate a spectral dataset with two class labels, antimicrobial active against MRSA and antimicrobial inactive against MRSA. Data distributed between these two classes were used to predict potential antimicrobial compounds. GC-MS spectra were pre-processed with the software tool XMCS to detect features and correct retention time drift. Linear discriminant analysis (LDA) was used to create a ranking classifier, which produced a list of compounds ordered by their significance in separating the antimicrobial active class from the antimicrobial inactive class. This list was compared to ranked compound lists generated by the software tool MetaboAnalyst and assessed by checking for the presence of compounds we experimentally found to have antimicrobial activity against MRSA. We found that a random forests analysis using MetaboAnalyst ranked 4 of 7 compounds experimentally observed to inhibit MRSA in the top 10 unique retention times, but also ranked two sucrose peaks in the top 10. LDA did not rank any known antimicrobial compounds in the top 10. Further validation is needed to properly assess the utility of these ranking systems, but the unidentifiable, highly ranked compounds from the random forests analysis merit further exploration as potential candidates for use as novel antimicrobials against MRSA


Poster 14

Comprehensive Analysis of Extractable from Rubber Stopper used in Medical Devices and Pharmaceutical Products

Kate Comstock

Thermo Fisher, CA

Comprehensive Analysis of Extractable from Rubber Stopper used in Medical Devices and Pharmaceutical Products

The use of multiple analytical techniques, HR-LCMS, GCMS, and ICPMS, is required for this analysis in order to identify all possible extractive species, and to ensure product safety and meet regulatory requirements.
The LC chromatographic separations were conducted on an Accucore C18 column (2.1x 100, 2.6 µm). The HRAM full scan and data dependent ms/ms data were acquired on Q Exactive.

Data were processed using software SIEVE and Mass Frontier (Thermo Scientific) for components extraction and structure elucidation. The preliminary data showed that the rubber stoppers have low level extractables.

Major component structures were confirmed through mzCloud high resolution spectral database searching. The putative structures of minor components were proposed based on ms/ms fragments using the “fragment and mechanism” feature in Mass Frontier.

GCMS analysis was performed in full scan data acquisition mode to identify unknowns. The peaks detected were identified using the NIST library search.

ICPMS analysis was carried out on iCAP Q ICP-MS with KED mode setting. USP <232> elements and other elements were determined and the results indicated that the extraction solutions were clean.

This study demonstrated a comprehensive workflow for extractable analysis using LCMS, GCMS, and ICPMS, employing data processing software SIEVE, Mass Frontier, and mzCloud database search.

This workflow utilizes unique features of each system to ensure the separation, identification, and characterization of complex extraction samples. With this workflow, confident and complete extractable analysis can be accomplished for various medical devices and pharmaceutical containers and packages.