Waters Medicinal Chemistry Applications Book

MEDICINAL CHEMISTRY APPLICATIONS BOOK Introduction...... 3

The Role of LC and MS in Medicinal Chemistry...... 5

SCREENING System Management for a High Throughput Open Access UPLC/MS System Used During the Analysis of Thousands of Samples...... 11

OpenLynx Open Access...... 15

CONFIRMATION New Tools for Improving Data Quality and Analysis Time for Chemical Library Integrity Assessment...... 23

PURIFICATION Scaling a Separation from UPLC to Purification Using Focused Gradients...... 29

Purification Workflow Management...... 33

Making a Purification System More Rugged And Reliable...... 39

Application of MS/MS Directed Purification to Identification of Drug Metabolites in Biological Fluids...... 45

Evaluating the Tools for Improving Purification Throughput...... 51

A Novel Approach for Reducing Fraction Drydown Time...... 57

PROFILING ProfileLynx Application Manager for MassLynx Software: Increasing the Throughput of Physicochemical Profiling...... 63

An Automated LC/MS/MS Protocol to Enhance Throughput of Physicochemical Property Profiling in Drug Discovery...... 65

OPTIMIZATION Synthetic Reaction Monitoring Using UPLC/MS...... 71

ACQUITY UPLC System: Time and Cost Savings in an Open Access Environment...... 73 The Role of Liquid Chromatography and Mass Spectrometry in Medicinal Chemistry

“ Medicinal chemistry is a scientific discipline at the intersection of chemistry and pharmacy involved with designing, synthesizing, and developing pharmaceutical drugs. Medicinal chemistry involves the identification, synthesis and development of new chemical entities suitable for therapeutic use.”

– Wikipedia.com

The objective of medicinal chemistry is to design and discover com- n In Screening, we will demonstrate the use of high UPLC pounds that offer the potential to become beneficial – and profitable throughput and fast-scanning MS to obtain high quality – therapeutic drugs. Confidently confirming the identity and quality and comprehensive data about compounds in the shortest of these new chemical entities is a major challenge, particularly possible time. when labs are asked to maximize throughput and efficiency – and to n For Compound Confirmation, we will show how an open access manage all the data generated by a variety of systems and users. interface, used with UPLC technology and advanced detection, Medicinal chemistry is also an iterative process that demands enables chemists with minimal instrument training to determine rapid turnaround times. High throughput liquid chromatography/ the identities of known compounds, to rapidly identify un- mass spectrometry (LC/MS), together with advanced data-handling knowns, and to characterize complex sample components. software, has become the standard technique for drug discovery compound identification and purification, addressing needs for high n In Purification, we provide several examples on how chemists throughput screening, optimization, and physicochemical property can use UPLC along with efficient time-saving techniques to profiling. dramatically increase throughput.

Waters UltraPerformance LC® (UPLC®) technology is providing a sea n In Compound Profiling, we illustrate an automated UPLC/MS/ change in capacity for medicinal chemistry labs. UPLC uses sub- MS protocol that not only allows for automated MS method de- two-micron column particle sizes to produce faster, more sensitive velopment and data acquisition, but also allows data generated and high-resolution separations. Our UPLC systems are available from multiple assays to be automatically processed by a single with fast-scanning detectors, both optical and mass, and can be processing method. easily controlled by software that facilitates sample analysis in n In Optimization, we will show how chemists were able to quickly open-access laboratory environments. and easily monitor their reactions, noting the relative amounts In this applications book, we look at a variety of system solutions of starting materials and products by using a walkup UPLC/MS that address the unique challenges of medicinal chemists in five system. key areas.

3 THE ROLE OF LIQUID CHROMATOGRAPHY AND MASS SPECTROMETRY IN MEDICINAL CHEMISTRY Darcy Shave, Paul Lefebvre, and Marian Twohig Waters Corporation, Milford, MA, U.S.

INTRODUCTION

Confirming the identity and quality of new chemical entities is a major challenge facing the pharmaceutical industry. Maximum efficiency is essential for laboratories challenged by throughput requirements and the management of data from a variety of systems and users.

Liquid chromatography with mass spectrometry has become the standard technique for confirming the identity and purity of drug discovery compounds to support high throughput screening (HTS), optimization, and physicochemical property profiling of these compounds. Medicinal chemistry is an iterative process and requires rapid Figure 1. The ACQUITY SQD with the Sample Organizer plus PDA and turnaround times. High throughput solutions together with advanced ELS detectors. data handling software must be employed.

In this application note, we look at various solutions, including Samples were analyzed on a Waters® ACQUITY UPLC® System sub-2 µm column particle sizes, fast scanning mass spectrometers, with a Sample Organizer. The column was an ACQUITY UPLC BEH and new software to assist the medicinal chemist in five key areas: C18 (1.7 µm, 2.1 x 50 mm) run at 30 °C. The injection volume was n Screening 5 µL. Compounds were separated using a generic water/acetonitrile n Confirmation gradient that was 1.1 min long. n Purification Detection was done with an ACQUITY UPLC Photodiode Array n Compound profiling (PDA), ACQUITY UPLC Evaporative Light Scattering (ELS), and SQ n Optimization Mass Detector with an ESCi® source for ESI/APCI switching. Plates were logged into and processed with the OpenLynx™ Open Access Application Manager for MassLynx™ Software.

METHODS AND DISCUSSION By using an ACQUITY UPLC System with the Sample Organizer, we Screening were able to analyze 3840 samples in under 7 working days on a single column. On a traditional HPLC system, this would take approxi- It is important to verify the identity and purity of a compound before mately 27 working days, assuming a 10-minute run time. early activity studies. Chemists need to be sure they have synthe- The ESCi source on the mass spectrometer allowed the chemist to sized the expected compound. Large numbers of compounds may be gather data in both electrospray and APCI (with positive/negative created, so it is necessary for this screening to be high throughput. switching) modes during the same injection. In this way, the maximum Because only a small amount of material is synthesized, the screening amount of data was generated with a minimal amount of sample. must also consume as little material as possible, while generating a diverse amount of information. The open access interface allowed the user to log in the sample plates while providing a minimal amount of information. A series of methods, each including gradient conditions, MS conditions, and processing parameters, was designed by the system administrator. The user simply chose a method from this list, imported their sample lists, and placed their microtitre plates in the indicated positions.

The samples were then analyzed and the data was processed. Once processing was finished, the data was automatically copied to a file storage PC. From here the users could do further processing, if desired. A report file was also generated from the processed file and converted to pdf. This facilitated storage of the results in a database.

Confirmation Figure 2. OpenLynx OALogin plate login wizard. Exact mass experiments permit elemental composition determi- nations of unknowns or confirmation of a suspected elemental composition. This allows the medicinal chemist to confirm identities A fast generic liquid chromatographic method was designed to provide of known compounds, to rapidly identify unknowns, and to character- excellent selectivity without compromising either chromatographic ize complex sample components. resolution or speed of analysis. To obtain such an analytical method, Samples were analyzed on an ACQUITY UPLC System. The column UPLC® in conjunction with oa-TOF MS detection was employed. With this analytical system, identification of the anticipated samples, was an ACQUITY UPLC BEH C18 (1.7 µm, 2.1 x 50 mm) run at 30 °C. The injection volume was 5 µL. Compounds were separated using a isomers, and possible impurities with mass accuracy deviations less generic water/acetonitrile gradient that was 1.1 min long. than 5 ppm from the actual were obtained using LockSpray™. With such high accuracy data, the calculation of elemental compositions Detection was done with an ACQUITY UPLC PDA and an LCT Premier™ for each of the analytes was possible. XE Mass Spectrometer with an ESCi source for ESI/APCI switching. Samples were logged into the system using OpenLynx Open Access Subsequent elemental composition results were produced using the and processed with MassLynx OpenLynx with i-Fit™ exact mass i-Fit algorithm, which takes into account the distribution of the spec- processing. tral isotopes for the compounds of interest and employs novel data interpretation to simplify results lists returned.

The Open Access interface allowed the medicinal chemist to log in the samples while providing a minimal amount of information. The results, including a pdf report showing the most probably elemental compositions, were then made available to the chemist.

6 Purification A rapid LC/MS method was developed for the analysis of a medicinal chemistry library. The MS data confirmed the presence of the target Having a pure building block is important for controlling the syn- compound and its retention time from a high resolution LC separation thetic reactions and successfully making a pure target. A pure target with a 1-minute cycle time. The retention time corresponded to a is critical for understanding the results of screening and building percent organic solvent at which the compound eluted. quality structure/activity relationship (SAR) information. Based on this correspondence, a focused purification method for a Reverse-phase HPLC has been successfully applied to the different 19 mm I.D. column with 5 micron particles was selected to maintain aspects of the medicinal chemist’s process. It is capable of purifying the analytical resolution. The isolated target was then separated by milligrams to multiple grams in a single system, and can be con- LC. The original analytical methodology was then used to determine figured to automatically process hundreds of samples. The results the new purity for each compound collected. can provide high purity and recovery of the desired compounds with minimal user intervention. By logging in their samples just once, the medicinal chemists were able to get a purified product along with reports showing the initial Samples were analyzed on a Waters AutoPurification™ System, and final purities. including a 2545 Binary Gradient Module, 2767 Injector, and Collector, and a System Fluidics Organizer (SFO). The compounds Compound profiling were purified on an XBridge™ Prep C ODB™ column (5 µm, 18 In an effort to avoid clinical failures, there is an emphasis across the 19 x 50 mm) run at room temperature. pharmaceutical industry on examining pharmacokinetic and safety Detection was done with 2996 PDA, ELS, and 3100 mass detectors. profiles earlier in the drug discovery process. Assays are developed Fraction collection and processing was done with the FractionLynx™ in order to select compounds with the highest probability of becom- Application Manager. Compounds were separated using 5-minute ing successful drugs based on preferred pharmacological properties. gradients that were chosen by the AutoPurify™ functionality of This step includes extensive testing for the absorption, distribution, FractionLynx. metabolism, excretion, and toxicity (ADMET) and physicochemical properties of a compound.

Samples were analyzed on an ACQUITY UPLC System with a Sample

Organizer. The column was an ACQUITY UPLC BEH C18 (1.7 µm, 2.1 x 50 mm) run at 30 °C. The injection volume was 5 µL. Compounds were separated using a generic water/acetonitrile gradient that was 1.1 min long.

Detection was done with an ACQUITY UPLC PDA, a ACQUITY UPLC ELS and a Quattro Premier™ XE Mass Spectrometer with an ESCi source for ESI/APCI switching. MS conditions were optimized using the QuanOptimize™ Application Manager. The samples were processed using the ProfileLynx™ Application Manager. Properties analyzed included solubility, logP, microsomal stability, and CHI.

Figure 3. MS and UV chromatograms showing targeted mass and impurities.

7 Optimization

Once a hit is generated through library screening, optimization of the compound of interest takes place. This step involves multiple repeti- tions of chemical modification of the hit to develop compounds with desired properties. Chemists need to know as soon as possible that these reactions are proceeding as desired.

Samples were analyzed on an ACQUITY UPLC System with a Sample

Organizer. The column was an ACQUITY UPLC BEH C18 (1.7 µm, 2.1 x 50 mm) run at 30 °C.

The injection volume was 5 µL. Compounds were separated using a generic water/acetonitrile gradient that was 1.1 min long.

Detection was done with an ACQUITY UPLC PDA, ACQUITY UPLC Figure 4. ProfileLynx browser showing results of solubility experiment. ELS and an SQ Mass Detector with an ESCi source for ESI/APCI switching. Single samples were logged into the system using OpenLynx Open Access and processed with the OpenLynx Early screening of physicochemical properties (PP) is an integral Application Manager. process for modern drug discovery. Typical PP profiling practices include properties such as solubility, stability (pH and metabolic), permeability, integrity, etc. The critical factor to consider in PP profiling is throughput. The bottlenecks to throughput include MS method optimization for a large variety of compounds and data management for the large volume of data generated.

An automated UPLC/MS/MS protocol was developed that not only allowed for automated MS method development and data acquisition, but also allowed data generated from multiple tests to be processed by a single processing method, all in an automated fashion. As a result, the physicochemical profiling process was significantly simplified and throughput increased.

The column manager bypass channel allowed users to easily switch to direct flow injection analysis for compound optimization without Figure 5. Chromatograms from various times during a 60-minute reaction. sacrificing one of the column positions. Chemists can choose the optimal conditions and chemistry for their compounds as the column manager is a thermostat-controlled oven with temperature regulation During the compound optimization stage of a discovery cycle, from 10 to 90 °C and has automated switching for four columns. medicinal chemists are not only interested in determining the key structural features responsible for activity and selectivity, but also what structural changes need be made to improve these characteristics. Because the reactions necessary to bring about these changes may take a long time, chemists need to be sure they are progressing as expected.

8 By using a walk-up UPLC/MS system, chemists were able to quickly n Purification: We were able to use analytical LC/MS data to tai- and easily monitor their reactions, noting the relative amounts of lor the purification method to maintain the analytical resolution. starting materials and products. They were also able to note the n Compounds profiling: The determination of physciochemical formation of any side products and make the necessary alterations to properties was simplified with the use of the ProfileLynx minimize these in their reaction protocol. Application Manager, which automated the calculations of solubility, logP, metabolic stability, and CHI. The combination of the Column Manager and QuanOptimize facilitated the development of optimal MS/MS method. CONCLUSION n Optimization: Chemists were able to quickly and easily log in their samples to determine the progress of the reaction. They We were able to increase throughput and data quality by were able to see the results of the analyses within minutes. combining UPLC with a variety of detection techniques and software solutions. n Screening: By combining the speed of the ACQUITY UPLC System with the capacity of the Sample Organizer, we were able to nearly quadruple the screening throughput of the lab, without sacrificing data quality. n Confirmation: With the Open Access interface, medicinal chemists were able to confirm the elemental composition of their compounds, with minimal instrument training. The i-Fit algorithm simplified the final exact mass determination by reducing the number of possible elemental formulas.

Waters, ACQUITY UPLC, UPLC, and ESCi are registered trademarks of Waters Corporation. AutoPurification, AutoPurify, FractionLynx, i-FIT, LCT Premier, LockSpray, MassLynx, ODB, OpenLynx, ProfileLynx, Quattro Waters Corporation Premier, QuanOptimize, XBridge, and The Science of What’s Possible 34 Maple Street are trademarks of Waters Corporation. All other trademarks are the Milford, MA 01757 U.S.A. property of their respective owners. T: 1 508 478 2000

©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720002099EN LB-KP www.waters.com SCREENING SYSTEM MANAGEMENT TOOLS FOR A HIGH-THROUGHPUT OPEN ACCESS UPLC/MS SYSTEM USED DURING THE ANALYSIS OF THOUSANDS OF SAMPLES Darcy Shave, Warren Potts, Michael Jones, Paul Lefebvre, and Rob Plumb Waters Corporation, Milford, MA, U.S.

INTRODUCTION

Many compound libraries contain compounds that were synthesized several years prior or obtained from outside resources. It is important that the expected composition of each compound be confirmed. LC/ MS has become the standard technique for confirming the purity and identification of a compound that has demonstrated activity in a biological screen.

If the library store is not routinely checked, false positives in an activity screen are highly possible. This will lead to wasted time, effort, and money on compounds that should not advance in the discovery process. Because these libraries may contain thousands, if The ACQUITY UPLC System with the ZQ Mass Detector for open access laboratories. not millions, of compounds, an Open Access UltraPerformance LC® (UPLC®)/MS system was investigated for high-throughput library quality control. EXPERIMENTAL

Enhancements to HPLC and LC/MS technologies have provided use- All experiments were conducted using the Waters® ZQ™ Mass ful tools to improve the throughput and accuracy of these assays. Detector, equipped with an ACQUITY UPLC® System with a Sample Throughput can be substantially increased with the use of UPLC/ Organizer, Photodiode Array (PDA) Detector, cooled Autosampler MS, which makes use of small column particles (sub-2 μm) and and Column Heater. The ZQ was equipped with an ESCi® source, high operating pressure (>10,000 psi). This can result in an up to running in the ES+ ion mode. The instrumentation was controlled 10-fold increase in throughput along with a three-fold increase in by MassLynx™ 4.1 Software with OpenLynx™ and OpenLynx Open sensitivity. Access Application Managers.

Due to the large number of samples analyzed and data generated Samples were run on a 1 min gradient from 5 to 95% organic at 0.8 during this testing, a new software package has been created that mL/min. The column was a 1.7 µm, 2.1 x 50 mm ACQUITY UPLC BEH facilitates administration of this Open Access system. It created new C18 Column. The PDA was set to analyze a wavelength range from project directories for the users and moved the resulting project data 210 to 400 nm. The mass detector analyzed a mass range from (such as raw data files) across the network as it was created. Data 100 to 500 amu with a dwell time of 100 ms and an interscan processing could then be done on a separate dedicated computer. delay of 50 ms. The software also monitored the instrument PC, providing on-the-fly Eight microtitre palates, each containing 96 pharmaceutical samples, information about its status and the status of its sample queue from were logged onto the system using OpenLynx Open Access. The first a centralized location. and last samples in each plate were used for quality control. RESULTS AND DISCUSSION Software administration tools

By using an ACQUITY UPLC System with the optional Sample The Open Access software allowed chemists to walk up to a terminal Organizer, we were able to analyze 3840 samples in under seven and log in samples onto an instrument, inputting the minimum of working days on a single column. On a traditional HPLC system, information needed for the sample run. It also allowed the system this would take approximately 27 working days, assuming a administrator to maintain control over the Open Access systems and 10-minute run time. to track the performance of each system. It facilitated batch processing and reporting of results. The Open Access interface allowed users to log in the samples while providing a minimal amount of information. A series of methods, The administrator selected the fields that appeared when remote users each including gradient conditions, MS conditions, and processing logged in samples. The administrator designated fields as manda- parameters, was designed by the system administrator. The users tory so that login would not proceed unless the remote users entered simply chose a method from this list, imported their sample lists, and values for these fields. They also defined upper and lower limits for placed their microtitre plates in the indicated positions. the values of numeric fields. In addition, the administrator defined the format for text that remote users entered in the text fields. The samples were then analyzed and the data was processed. Once processing was finished, the data was copied to a file storage PC. The Open Access Toolkit (OAToolkit) service ran on the Acquisition From here the users could do further processing if desired. As well, PC and copied open access users’ batch files and raw data to remote a report file was generated from the processed file and converted to locations once their samples were run. The information about these .xml format. This facilitated storage of the results in a database. users, and the locations to where their data was to be sent, is contained within the administration tool. This information is uploaded to Instrumentation the service on the Acquisition PC. Throughput was increased by using UPLC. This technique made use The illustration in Figure 1 and following procedure describe the of 1.7 μm column particles and high operating pressure (12,000 order of events during typical operation. psi). These properties resulted in a five-fold increase in throughput. Sensitivity was not investigated. 1. The administrator uses the Administration Tool to create a user.

Due to the large number of samples being run, an ACQUITY UPLC 2. The administrator uses the Administration Tool to add extra Sample Organizer was also used. This thermally-conditioned sample information about the OALogin user, for example, that the raw storage compartment extended the capacity of the system by adding data of any of the user’s samples should be moved to the File space for seven deep-well microtitre plates (or 21 shallow-well plates). Storage PC whenever a user’s sample is processed. Total sample capacity was increased from 192 samples (two plates) 3. The administrator uploads the user information to the OALogin to 768 samples (eight plates) when using 96-well plates. If using PC. This adds the user’s name to the drop-down list in the 384-well plates, maximum capacity would be 8064 samples. login screen on the OALogin PC. An added advantage of the Sample Organizer in an open access 4. The administrator uploads the user information to the OAToolkit environment is the ability to add samples to the system without service on the Acquisition PC. The service now contains the pausing the sample queue. When the door to the Sample Manager instructions of how to proceed if the OALogin user logs in a batch. is opened, any movement –whether of the sample plate or of the needle – is paused for safety. This pause does not occur when loading 5. The OALogin user logs in a sample using the OALogin terminal the Sample Organizer. as normal.

6. OALogin logs the sample with MassLynx.

12 7. When MassLynx has finished running the sample, the OAToolkit service reads the batch file (.olb), and registers that it is from a recognized user.

8. The OAToolkit service moves the raw information to the specified location on the File Storage PC.

7. 8.

Acquisition PC File Storage PC Mass Spectrometer

4. 6.

1. 5. 3. Figure 2. OpenLynx browser report. 2.

OAToolkit OAToolkit Administration PC Administration PC The browser report was created in the report folder of the current project. A secondary report location could have been specified, but was not. The toolkit service also allowed for a copy of the report to be Figure 1. Data from the mass spectrometer is captured by the Acquisition PC, then is managed by the system administrator or accessed by the individual user sent over the network to another location. That location was specific via the OALogin tool. The raw data is also backed up to a File Storage PC. to each user – a folder on their office PCs. The users no longer had to access the Acquisition PC to view their reports. In addition, the raw data folders were moved across the network to each user’s PC and Reporting the users were able to reprocess it with a process-only version of MassLynx. The Open Access software allowed the administrator to define how samples were processed. Once all the data for a sample set had Finally, the OAToolkit service was used to automatically convert been collected, the OpenLynx Application Manager automatically the browser reports to .xml format. This was accomplished using the processed the data and created an OpenLynx Browser report (.rpt). included .xml import and .xsl export schema. This data can then be easily incorporated into a database or shared with colleagues. The browser report (Figure 2) presented a summary of results as a color-coded map (found/not found/tentative) for easy visualiza- System monitoring tion of analysis results. Users accessed and reviewed the data by On the Administration PC, the Remote Status Monitor (RSM) moni- simply pointing and clicking on the sample location of interest. tored the status of the Open Access Acquisition PC, along with other Chromatograms, spectra, sample purity, peak height, peak area, Acquisition PCs on the network and wrote that monitoring information retention time, and other information can easily be reviewed within to an .xml file. The information could then be read and interrogated the browser. remotely in a browser (Figure 3).

13 CONCLUSION

Waters Open Access systems give chemists the ability to analyze their own samples close to the point of production by simply walking up to the LC/MS system, logging in their samples, placing their samples in the system as instructed, and walking away. As soon as Figure 3. Status of the Open Access Acquisition PC. the analysis is completed, sample results are emailed or printed as desired. System configuration and setup is enabled through a More detailed information about an instrument can be displayed system administrator who determines login access, method selec- by clicking anywhere in the instrument row (Figure 4). tion, and report generation.

OpenLynx OAToolkit enables administrators to manage open access users from a central point, assigning detailed configuration information and attributes for these users, and then exporting these details to multiple OALogin PCs and Acquisition PCs. OpenLynx OAToolkit also enables administrators and users to remotely monitor the status of Acquisition PCs.

Figure 4. Detailed view of instrument status.

Waters, ACQUITY UPLC, ESCi, UltraPerformance LC, and UPLC are registered trademarks of Waters Corporation. MassLynx, Waters Corporation OpenLynx, ZQ, and The Science of What’s Possible are 34 Maple Street trademarks of Waters Corporation. All other trademarks Milford, MA 01757 U.S.A. are the property of their respective owners. T: 1 508 478 2000 ©2006-2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001482EN LB-KP www.waters.com OPENLYNX OPEN ACCESS

OVERVIEW OpenLynx Open Access offers comprehensive capabilities: n Simplified sample submission process – A single page login Maximum efficiency is essential for LC/MS labs challenged by or a step-by-step, wizard-enabled process allows users to enter throughput requirements and the management of data from a vari- their name and sample information, and select pre-determined ety of systems and users. Analyzing routine samples and returning experimental methods and processing criteria the results to chemists can easily consume an analyst’s entire n Exact mass measurement utilization – For use with the day, leaving them with little time to focus on tasks that require appropriate mass spectrometers their expert attention. Walk-up open access systems allow chem- n Summary report generation – Reports are automatically ists to analyze their own samples, freeing up analysts’ time for printed, emailed, and viewed via the OpenLynx browser, more challenging analyses without compromising the quality of the containing sample found/not found information, purity, final results. probable elemental composition (with exact mass MS), The Waters® OpenLynx™ Open Access Application Manager for chromatograms, and spectra MassLynx™ Software offers the power of chromatography and mass n Walk-up optimization of MS/MS methods and quantification spectrometry to chemists who are not analytical instrumentation of compounds of interest – Combines OpenLynx Open Access specialists. To minimize the learning curve for instrument operation, with QuanOptimize™ and QuanLynx™ Application Managers OpenLynx Open Access leads chemists through sample submission, n Advanced search – Spectral library generation and searching method selection, and reporting options. The system is maintained n Automation of routine system administration tasks – by a system administrator who predefines the system configuration, Through the use of OpenLynx Open Access Toolkit (OAToolkit) available experimental methods, processing criteria, and reporting options. By allowing chemists to submit their own samples, routine analyses can be performed more efficiently, leaving instrumentation experts more time to focus on advanced analyses. SOFTWARE SETUP Defining parameters

OpenLynx Open Access allows remote users to run samples on INTRODUCTION the acquisition computer. For OpenLynx Open Access users to be successful, the administrator defines (via the OpenLynx method) Open access LC/UV, LC/MS, LC/MS/MS, and GC/MS the sample information that users must provide when running The OpenLynx Open Access Application Manager is designed to allow samples. An intuitive OALogin setup wizard simplifies the system chemists to walk up to a terminal and log in samples onto an instru- configuration and administration workspace to include only the ana- ment, while inputting the minimum of information needed for the lytical features the administrator uses. sample run. OpenLynx Open Access allows the system administrator to maintain control over the open access systems and to track the performance of each system. It also facilitates batch processing and reporting of results. Figure 1. OpenLynx method showing some of the OpenLynx Open Access input fields. Figure 2. Administrator-set OpenLynx Open Access options.

The administrator selects the fields that appear when remote users log Setting file options in samples using OpenLynx Open Access via the Walk-up tab of the The administrator sets several file options. These include specifying OpenLynx method (Figure 1). They can designate fields as mandatory the location where the OpenLynx methods, OpenLynx status file, and so that login will not proceed unless the remote users enter values for HPLC files are located. The administrator can set which methods are these fields. They can also define upper and lower limits for the value visible to users, along with the format needed for the text fields. of numeric fields. In addition, the administrator can define the format for text that remote users enter in text fields. Configuring quality control runs

Setting options for users The administrator can configure OpenLynx to check that the LC and MS instrumentation are working correctly, thus ensuring the consis- Using the administrator mode of OpenLynx Open Access, the admin- tency of the data. The quality control feature (Figure 3) allows users istrator defines how users login samples via a number of options to run a standard and have it compared to the results of the same (Figure 2). Login setup ranges from changing the window appear- standard that was run at an earlier time. Values that can be used ance to allowing users to create their own user name. Notification of to confirm system operational performance include peak retention users via email can be enabled, as can barcode support. OALogin can time, peak area, the presence of specific masses or wavelengths, and be configured for use with either OpenLynx (sample processing) or spectral intensity. AutoPurify™ (fraction processing). Figure 4. OpenLynx Toolkit Administrator Tool.

Figure 3. OpenLynx Open Access quality control options.

Before a QC comparison can be run to check the system, there must be an OpenLynx method that contains the expected results from a standard. The QC run acquires data from a sample with a known retention time and peak intensity and then compares the results to the values defined in the OpenLynx method. Figure 5. Remote Status Monitor.

OpenLynx Open Access Toolkit (OAToolkit)

OpenLynx OAToolkit allows the creation and administration of The OpenLynx OAToolkit includes the following key features: OpenLynx Open Access users. It can push user information to n Administration Tool (Figure 4) – Enables an administrator to OpenLynx Open Access PCs across the same network, as well as create and manage all OpenLynx Open Access users from a gather existing OpenLynx Open Access user information from single PC, and replicates that information to multiple OpenLynx OpenLynx Open Access PCs. It can create new project directories for Open Access PCs and Acquisition PCs the OpenLynx Open Access users and can move the resulting project n OAToolkit Service – Runs in the background on one or more data (such as raw data files) as it is created. The software can monitor Acquisition PCs, monitors sample batches submitted by numerous instrument PCs, providing on-the-fly information about OpenLynx Open Access users that were uploaded from the their status as well as the status of their batch queues – all from Administration Tool a central location. It ensures confidence in analytical results with n Remote Status Monitor (Figure 5) – Enables any user to password protection for open access users. monitor the status of Acquisition PCs and their batch queues from a single PC Additionally, the OpenLynx OAToolkit Service: n Relocates data produced during the processing of an OpenLynx Open Access user’s batch of samples n Creates new project folders in which to store the processing data on a timed basis n Converts report files to different formats (XML, HTML, or text)

LOGGING SAMPLES

Login samples window Figure 6. OpenLynx Open Access window. Running samples using OpenLynx Open Access (Figure 6) involves entering sample information to correctly identify the samples and loading the samples into the autosampler. The methods available to the users depend on selections made by the administrator.

If the administrator enables user passwords (using OpenLynx OAToolkit), the user must enter their designated password before they can login samples (Figure 7). If they enter an incorrect password, an error message appears and they cannot continue until the correct password has been entered.

Single-page log-in vs. wizard

OpenLynx Open Access displays the wizard for sample login by default. However, the administrator can allow OpenLynx Open Access Figure 7. Entering user password. users to use a single-page dialog box (Figure 8) for “single shot” samples. Users can enter multiple samples in this way. OpenLynx Open Access views the samples logged in as a single job. PROCESSING SAMPLES

Processing data automatically

The administrator determines how OpenLynx processes the Open Access results. To configure OpenLynx Open Access to process data automatically, the administrator must create an OpenLynx method that defines the processing parameters.

The administrator must define the integration parameters for the type of data they want to process: n MS+ data – For positive ions (total ion chromatogram (TIC), base peak intensity (BPI), and mass chromatograms) n MS– data – For negative ions (TIC, BPI, and mass chromatograms) n Analog data – For up to four channels of analog chromatograms

Figure 9. With the wizard, walk-up users enter their name, choose a method, n DAD data – For total absorbance chromatogram (TAC), BPI, enter sample information, and place the sample in the autosampler. and wavelength chromatograms

Specifying how peak detection occurs involves selecting the integration algorithm and parameters that control peak detec- The single-page login contains most of the selections on the tion, enabling smoothing (if desired), and setting the smoothing wizard pages (Figure 9) necessary to schedule samples. The benefit parameters and setting threshold values. of the single-page login is the speed of entering information for a single sample in a single dialog box, rather than through a wizard. This wizard is beneficial when logging in larger sample sets.

Loading samples into the autosampler

There are two ways to load samples into the autosampler. The system administrator designates each plate in the autosampler as either “single shot” or “whole plate” login. If a plate is designated for single shot login, the user enters data for their samples manually or imports data from a tab-delimited text file. OpenLynx assigns available positions for the samples on existing plates. If a Figure 10. Chromatogram integration window. plate is designated for whole plate login, the user prepares data in a spreadsheet or as a text file and imports it into OpenLynx Open When setting the integration and peak detection parameters (Figure Access. This is useful if the user needs to run a large number of 10), the administrator can specify which integration algorithm samples in one run. OpenLynx reserves the entire plate for samples (standard or ApexTrack™) to use; how the baseline will be treated and the user selects the sample locations. for valleys, peak tailing, and drift; and how peak separation for fused Typically, a system with multiple plates will have both single shot and whole plate login available. peaks and shoulders will be handled. By enabling smoothing, Using Open Access quantitation, OpenLynx Open Access users noise will be decreased by filtering data points. Smoothing types can quantify the results as data are acquired. The processing steps include Savitzky-Golay and mean. The threshold values are set available include: for one or more of the four threshold parameters: relative and n Integrating samples absolute height and relative and absolute area. This option is n Quantitating samples used to remove peaks whose height or area is less than a specified n Calibrating standards percentage of the highest peak. Using QuanOptimize with OpenLynx Open Access In addition to acquiring and processing data, quantitation and optimi- The optional QuanOptimize optimizes the acquisition and quantitation zation can be performed through OpenLynx Open Access. parameters for a particular experiment. Open Access QuanOptimize Performing quantitation (Figure 12) generates MS and MS/MS parameters by optimizing the cone voltage, parent ion, and collision energy parameters. Open Access quantitation is a way for the user to run quantitation QuanOptimize then takes these MS methods and performs automated analysis through OpenLynx Open Access (Figure 11). OpenLynx acquisition and processing using processing methods developed on stores the conditions required for a particular quantitation analysis the fly. It can quantify these results using specified methods. This in an OpenLynx method. OpenLynx Open Access users select the technique is useful for high throughput screening. OpenLynx method during login.

Figure 11. Open Access quantitation parameters. Figure 12. Open Access QuanOptimize parameters. REPORTING The OpenLynx browser presents a summary of results as a color- coded (found/not found/tentative) map for easy visualization of Results reporting analysis results. Chemists can access and review the data supporting Reporting in Open Access systems is facilitated by the OpenLynx any found/not found/tentative assignment by simply pointing and Application Manager. OpenLynx can report results using a flexible clicking on the sample location of interest. Chromatograms, spectra, array of printed reports or through a results browser. sample purity, peak height, peak area, retention time, and other information can easily be reviewed within the browser. The standalone OpenLynx browser (Figure 13) is an interactive tool for viewing OpenLynx results and can be run on any windows PC without requiring a full MassLynx installation. Chemists can use the browser on their desktop PC to view the results (.rpt file format) that had been automatically emailed to them at the end of OpenLynx processing.

Figure 13. OpenLynx browser. Printing and distributing reports CONCLUSION

OpenLynx creates an OpenLynx browser report file (.rpt) after The OpenLynx Open Access Application Manager provides it finishes a run and processes the data. This file resides in the comprehensive, easy, and flexible open access walk-up LC/UV, OpenLynx Open Access\Reportdb folder. The file is named with the LC/MS, LC/MS/MS, and GC/MS systems operation management job number followed by the extension .rpt when the user logs in to for laboratories that have chemists with varying levels of instru- OpenLynx. OpenLynx report files may be exported in .txt, .tab, .csv, mental analysis experience. With customizable batch processing and .xml formats. and results review to support the large amounts of data resulting from high throughput analyses, a highly productive environment is The administrator can configure OpenLynx Open Access so remote ensured for high-volume laboratories. users can find the reports that OpenLynx generates after running samples. Information such as where to store reports and what print report format to use can be specified.

Waters is a registered trademark of Waters Corporation. MassLynx, QuanOptimize, QuanLynx, ApexTrack, AutoPurify, Waters Corporation OpenLynx, and The Science of What’s Possible are trademarks 34 Maple Street of Waters Corporation. All other trademarks are the property Milford, MA 01757 U.S.A. of their respective owners. T: 1 508 478 2000 ©2006-2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001594EN LB-KP www.waters.com confirmation NEW TOOLS FOR IMPROVING DATA QUALITY AND ANALYSIS TIME FOR CHEMICAL LIBRARY INTEGRITY ASSESSMENT Marian Twohig, Paul Lefebvre, Darcy Shave, Warren Potts, and Rob Plumb Waters Corporation, Milford, MA, U.S.

INTRODUCTION

The identity and purity of a candidate pharmaceutical is critical to the effectiveness of the drug screening process. LC/MS is employed extensively in drug discovery in order to exclude false positives and maintain the high quality of the product. This process can be time consuming and can potentially delay the progression of a drug through the discovery process.

Thus, sample throughput is a critical issue in moving compounds from the hit to lead status. UltraPerformance LC® (UPLC®) lever- ages sub-2 µm LC particle technology to generate high efficiency faster separations.

When a photodiode array/evaporative light scattering/mass spectrometry (PDA/ELS/MS) detection scheme is used in conjunction Figure 1. The ACQUITY SQD for open access. with multiple-mode ionization, the potential for peak detection is greatly improved. Pharmaceutical chemical libraries often contain a great diversity of small molecules to cover a broad range of biological targets.1 In this environment, the ability to obtain infor- EXPERIMENTAL mation pertaining to multiple MS acquisition modes, in addition to PDA and ELS, in a single injection is invaluable. LC conditions

® ® Open Access software offers the power of chromatography and LC system: Waters ACQUITY UPLC System Column: acQUITY UPLC BEH C Column mass spectrometry to chemists who are not analytical instru- 18 mentation specialists. It allows them to quickly and easily know 2.1 x 30 mm, 1.7 µm what they’ve made and allows the experts to work on the difficult Column temp.: 50 °C analytical problems. Sample temp.: 8 °C Injection volume: 2 µL An Open Access UPLC/MS system was investigated for high Flow rate: 800 µL/min throughput library QC. In this application note, we describe some Mobile phase A: 0.1% Formic acid in water of the enhancements to LC and LC/MS technologies that have Mobile phase B: 0.1% Formic acid in acetonitrile generated useful tools that improve the throughput and accuracy Gradient: 5 to 95% B/0.70 min of these assays. MS conditions Sample login

MS system: Waters SQ Detector OpenLynx™ Open Access Application Manager for MassLynx™ Ionization mode: eSI positive/ESI negative, Software is designed to allow chemists to walk up to a terminal and multi-mode ionization log in samples while entering the minimum information required to Capillary voltage: 3.0 KV run the samples. A series of methods, each including gradient and Cone voltage: 20 V MS conditions as well as processing parameters, are initially set up Desolvation temp.: 450 °C by the system administrator. The users choose an appropriate method Desolvation gas: 800 L/Hr from the list, importing their sample lists and placing their samples Source temp.: 150 °C in the position designated by the software. Desired sample analysis Acquisition range: 100 to 1300 m/z is then performed by the configured system. The single page login Scan speed: 2500, 5000, and 10,000 amu/sec window can be seen in Figure 2.

Note: A low volume micro-tee was used to split the flow to the ELS and SQ.

ELS conditions

Gain: 500 N2 gas pressure: 50 psi Drift tube temp.: 50 psi Sampling rate: 20 points/sec

PDA conditions

Range: 210 to 400 nm Sampling rate: 20 points/sec Figure 2. OpenLynx Open Access single page login.

RESULTS AND DISCUSSION Open access system

Maximum efficiency is essential for labs challenged by throughput Chromatographic separations were carried out using the ACQUITY® requirements and the management of data from multiple systems SQD System coupled to detectors specialized for UPLC separations: and users. The Waters Open Access suite of software streamlines the single quadrupole SQ Mass Detector, and PDA and ELS detectors the integration of analysis with data acquisition, processing, and that provided simultaneous signal collection. For additional flexibil- reporting. ity, the UPLC system was configured with a Sample Organizer and a Column Manager. The sample capacity of the system totals twenty two The system and software are initially configured by a system admin- 384-well plates, for 8448 library samples in total. This extends the istrator who defines login access, method selection, and reporting overall walk-away time for the system. The column manager allows schemes. This allows users to analyze their own samples with mini- four UPLC columns to be installed, heated, and switched into line mal intervention from analytical support.

24 based on the method requirements. This allows the chemist to take The total cycle time of the method was 1 minute 20 seconds, advantage of the broad range of stationary phases that encompass facilitating increased sample throughput while still allowing gener- compound types, ranging from very hydrophilic to very lipophilic. ous washing steps to prevent sample-to-sample memory effects. Using a flow rate of 800 µL/min and a 2.1 x 30 mm column clears Sample analysis 9 column volumes/min. Samples were analyzed using gradients less than one minute in The spectral data quality of scanning experiments carried out from length with a flow rate of 800 µL/min. When analyzing the narrow 2500 to 10,000 amu/sec were found to be comparable, thus provid- peaks generated by the UPLC/MS system, the data collection rate can ing confidence that operating at these rapid data collection rates compromise the number of points across the LC peak, resulting in a does not compromise the spectral data quality. Figure 4 shows a poor definition of the eluting peak and hence inaccurate results. comparison of an acquired spectrum with a software generated isoto- The ability of the MS system to collect data at a high scan speed, pic model. Isotope ratios of data collected at 10,000 amu/sec were 10,000 amu/sec, greatly improves chromatographic peak defini- within 1% of the isotopic model, again ensuring data fidelity is not tion. This in turn facilitates the acquisition of a large number of compromised. individual acquisition modes in one run while maintaining adequate In addition to obtaining mass confirmation by multiple MS modes, it peak characterization. is possible to add PDA and ELS detectors to obtain auxiliary informa- As can be seen from the data displayed in Figure 3, the result of tion. A single run can then provide UV spectral information and an operating at lower data collection rates can compromise the chro- estimation of compound purity at low wavelengths. matographic resolution. To maintain chromatographic integrity, it is therefore advantageous to be able to scan at elevated scan speeds.

0.30 319 2500 amu/sec 100 0.53

0.26 C17H20N2CIS

% Acquired Spectrum 10,000 amu/sec %

0 321 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70

100 0.30 5000 amu/sec

0 0.53 0.26 %

100 319

0 C17H20N2CIS 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 Isotope Model

100 0.30 10,000 amu/sec % 0.25 0.53 321 %

0 Time 0 m/z 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 305 310 315 320 325 330

Figure 3. Chromatograms shown at 2500, 5000, and 10,000 amu/sec. Figure 4. Spectrum for isotope model and for acquired spectrum.

25 ELS detection is an alternative to UV detection, and does not depend 0.31

- on the presence of a chromaphore. ELS detection works by measur- 1.0e-1 PDA

AU - 0.0 ing the light scattered from the solid solute particles remaining after 0.10 0.20 0.30 0.40 0.50 0.60 0.70 nebulization and evaporation of the mobile phase. Chromatograms 0.30 0.54 illustrating the use of triple detection (PDA/ELS/MS) are shown in 40.000 0.24 ELS U 20.000 Figure 5. The signal from an ELS detector can give a tentative estima- LS 0.000 tion on the relative quantities of the components present. It has been 0.10 0.20 0.30 0.40 0.50 0.60 0.70 known to give rise to similar responses for related compounds.2 0.31 100 0.54 APcI+ % The chromatographic peak widths of the MS and ELS increased by 0 0.10 0.20 0.30 0.40 0.50 0.60 0.70 25 to 30% when compared with the PDA trace. This can be attributed APcI- to the use of a low volume micro-tee after the PDA. 0.180.24 0.32 0.56 0.64 0.10 0.41 0.48 % Data processing 12 0.10 0.20 0.30 0.40 0.50 0.60 0.70 As soon as the analysis is complete, data is automatically pro- 0.54 cessed and a sample report is generated. Reporting in Open Access ESI- % systems is facilitated by the OpenLynx Application Manager. 16 0.10 0.20 0.30 0.40 0.50 0.60 0.70 OpenLynx can report results using printed reports or through the OpenLynx browser. The browser presents a summary of the results 0.31 0.54 100 0.25 as a color coded (found/not found/tentative) map for clear interpre- % ESI+ 0 Time tation of the results. Chromatograms, spectra, sample purity, peak 0.10 0.20 0.30 0.40 0.50 0.60 0.70 height, peak area, retention time, and other information can easily Figure 5. UPLC/PDA/ELS/MS with multi-mode ionization. be viewed by the browser. The OpenLynx browser, shown in Figure 6, displays the results for the entire 384-well plate. The report can automatically be emailed, converted to pdf, or printed as desired.3

The OpenLynx OAToolkit facilitates an even easier administration of an open access system, automating many of the system management tasks carried out by a system administrator. The software also remotely monitors the status (via the Remote Status Monitor module) of one or more acquisition PCs and writes monitoring information to an XML file. The status summary page opens in the browser and contains a list of acquisition PCs, and the number of samples pending in the queue.4 This allows the chemist to select the instrument with the shortest wait time, again increasing productivity.

Figure 6. The OpenLynx browser.

26 CONCLUSION Open Access gives the chemist a walk-up system that is flexible for analytical data acquisition. It runs as a complete system, from sample It is important to verify the identity and purity of a compound before introduction to end results. early activity studies. Chemists need to be sure they have synthesized the expected compound. Because large numbers of compounds may The use of the fast-scanning MS along with the throughput of UPLC be created, it is necessary for this screening to be high throughput. technology and remote status monitor software allows the chemist to And because only a small amount of material is synthesized, the obtain high quality comprehensive data about their compounds in the screening must also consume as little material as possible, while shortest possible time. This combined with intelligent open access generating a diverse amount of information. software allows informed decisions to be made faster, thus supporting the needs of the modern drug discovery process. The described system and software combination can autonomously evaluate large numbers of samples with a cycle time of 1 minute and 20 seconds. Data can then be automatically processed and a sum- References mary report can be generated. The scan speed capabilities of Waters 1. Mike S. Lee, LC/MS Applications In Drug Development, Wiley-Interscience ACQUITY SQD System make it possible to better characterize narrow Series on Mass Spectrometry. 2002; (Chapter 6) 96-106. chromatographic peaks. This has become a necessity since the advent 2. kibbey, C.E. Mol. Diversity. 1995; I: 247-258. of sub-2 µm particle technology where chromatographic peaks can be 3. darcy Shave, OpenLynx Open Access, Waters Application Library. 2006: 1 second wide or less. 720001594EN. 4. darcy Shave, OpenLynx OAToolkit for Open Access Systems, Waters Signals from auxiliary detectors such as PDA and ELS can be col- Application Library. 2006: 720001469EN. lected simultaneously. Together with the MS data, they provide important information relating to purity and an estimation of the relative quantities of the components present.

Waters, ACQUITY, ACQUITY UPLC, UltraPerformance LC, and UPLC are registered trademarks of Waters Corporation. Waters Corporation MassLynx, OpenLynx, and The Science of What’s Possible are 34 Maple Street trademarks of Waters Corporation. All other trademarks are Milford, MA 01757 U.S.A. the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720002257EN LB-KP www.waters.com purification SCALING A SEPARATION FROM UPLC TO PURIFICATION USING FOCUSED GRADIENTS Ronan Cleary, Paul Lefebvre, and Marian Twohig Waters Corporation, Milford, MA, U.S.

INTRODUCTION

Purification laboratories face many of the same challenges that their counterparts in analytical laboratories face: the need to increase throughput and efficiency without sacrificing quality and quantity. Successful performance of a purification lab is measured in the ability to produce pure fractions in sufficient quantities in a timely manner.

UltraPerformance LC® (UPLC®) has been widely accepted by chromatographers because of the improvements over HPLC in sensitivity, resolution, and speed of separations. Now scientists are beginning to explore the use of this technology in the sample screening process as a screening tool to evaluate samples prior to purification. Figure 1. The mass-directed AutoPurification System. A typical run time for analytical screening in a preparative lab is 10 minutes. By capitalizing on the efficiency of UPLC, a 10-minute run time can be shortened to as little as 1 minute. This offers sub- UPLC conditions stantial time savings enabling for greater capacity, but also fits into the “fail fast and fail cheap” motto adopted by many pharma- LC system: Waters® ACQUITY UPLC® System with ACQUITY ceutical companies. UPLC Photodiode Array (PDA) Detector Column: acQUITY UPLC BEH C , 1.7 µm, 2.1 x 50 mm This application note will discuss the use of focused gradients to 18 Injection volume: 2.0 µL maintain selectivity and resolution and to allow UPLC screening Flow rate: 0.8 mL/min, 2.1 x 50 mm to be applied to preparative samples. This will offer the substantial Mobile phase A: 0.05% Formic acid in acetonitrile time savings associated with UPLC to customers in the preparative Mobile phase B: 0.05% Formic acid in water environment. Gradient: Generic 5 to 95% over 2 minutes Focused Gradient

HPLC conditions EXPERIMENTAL LC system: Waters AutoPurification™ System A standard solution of pharmaceutical-like compounds was Column: Waters XBridge™ Prep OBD™ C18, prepared to simulate the conditions under which many purification 5 µm, 19 x 50 mm systems operate. Waters XBridge C18, 5 µm, 4.6 x 50 mm Injection volume: 200 µL Mobile phase A: 0.05% Formic acid in acetonitrile Mobile phase B: 0.05% Formic acid in water Flow rate: 22 mL/min Gradient: 0 to 0.25 min, 2% B to initial % B 0.48 0.25 to 1.61 min, initial % B to end % B 0.27 1.61 to 1.86 min, end % B to 95% B 2.0 0.34 0.51 1.86 to 2.71 min, 95% B AU 2.71 to 2.72 min, 95% B to 2% B 1.0 0.13 0.23 MS conditions 0.0 Time MS system: Waters 3100 Mass Detector 0.00 0.20 0.40 0.60

Ionization mode: positive Figure 2. ACQUITY UPLC analytical separation. Switching time: 0.05 sec Capillary voltage: 3 Kv Cone voltage: 60 V 0.64 Desolvation temp.: 350 °C

Desolvation gas: 500 L/Hr 4.0e+1 3.88 Source temp.: 300 °C 1.61 AU Acquisition range: 150 to 700 amu 2.0e+1 2.31 4.17 Acquisition rate: 5000 amu/sec

0.0 Time 2.00 4.00 6.00 8.00

Figure 3. Direct scale-up maintains resolution and selectivity, with a run time of RESULTS AND DISCUSSION 8 minutes.

In order to maintain the selectivity and resolution achieved by analytical analysis, the overall cycle time of a preparative analysis must be increase almost nine fold.1 This long cycle time In a preparative environment, where the compound of interest is is not practical for most separation scientists. Therefore, we look being isolated from the other components in the sample, retaining to focused gradients to maintain selectivity and resolution in analytical resolution is not as important as isolating and collecting 2 UPLC screening. the compound of interest.

The UPLC separation of the sample shows the compound of A set of focused gradients can be created based on the relationship interest eluting at 0.48 min, and is partially resolved from the between percent composition and retention time. The system dwell 3 peak at 0.51 min. time is used to determine that relationship.

The separation is first directly scaled to a 19 x 50 mm XBridge Prep Here, in the analytical screen the mobile phase is 2% organic solvent at 0.17 minutes and 17.5% at 0.295 minutes, and so a series of OBD C18 Column. The XBridge chemistry is built on the same second- generation bridged ethyl hybrid (BEH) particle as the ACQUITY gradients can be created. UPLC BEH chemistry, in order to maintain the selectivity and The theory behind the focused gradients is the same for HPLC resolution of the analytical analysis. To maintain the resolution and and for UPLC, but the time window for the UPLC gradient is selectivity, the overall cycle time must be increased over nine fold. much smaller.

30 Based on Table 1, method C is selected to isolate the compound that eluted at 0.48 min in the UPLC analysis. Using the focused gradient, the separation and isolation of the compound was carried out in 3 minutes.

Method Time (min) Time (min) % B start % B end A 0.17 0.295 2 17.5 B 0.295 0.42 17.5 33 C 0.42 0.545 33 48.5 D 0.545 0.67 48.5 64 E 0.67 0.795 64 79.5 F 0.795 0.92 79.5 95 Figure 5. AutoPurify processing report showing the color coded purity and Table 1. UPLC retention time windows and corresponding focused preparative found/not found of a 348-well plate. gradient composition.

0.83 Focused library purification AutoPurify automatically selects the samples requiring purification 1.0e+2 2.01 and the corresponding focused preparative method. 0.63 AU 5.0e+1 1.08 2.24

0.0 Time 1.00 2.00 3.00

Figure 4. Separation of the compound of interest using a 3-minute focused gradient.

UPLC library purity screening

This same methodology can be applied to the purity screening and purification of a large sample library. The ACQUITY UPLC System’s large capacity (22 384-well plates) and the rapid analysis cycle time provide the ideal tool for high throughput library screening. Data is Figure 6. AutoPurify processing of the UPLC screening library. processed and handled using AutoPurify™, part of the FractionLynx™ Application Manager™.4

31 UPLC fraction analysis CONCLUSION n Scale-up from UPLC to preparative HPLC in an efficient manner The substantial time savings associated with analytical screening can is possible with the use of focused gradients. be magnified by incorporating UPLC into the analysis of the collected n The efficiency of UPLC can be carried through to purification, fractions. The collected fractions are analyzed to determine the new offering a substantial increase in throughput and productivity. sample purity, and sample lists are automatically generated for each n The AutoPurify capabilities of FractionLynx allows for step of the process. By incorporating fraction analysis by UPLC into automation from the initial UPLC QC, through purification, the workflow, the efficiency of the lab is further increased. to UPLC fraction analysis. n AutoPurify is also capable of automatically selecting a focused preparative gradient based on the analytical results, giving better quality purification and eliminating the need for expert manual invention.

References

1. Xia F, Cavanaugh J, Diehl D, Wheat T. Seamless Method Transfer from UPLC Technology to Preparative LC, Waters Application Note. 2007; 720002028EN.

2. cleary R, Lefebvre P. The Impact of Focused Gradients on the Purification Process, Waters Application Note. 2007; 720002284EN.

3. Jablonski J, Wheat T. Optimized Chromatography for Mass Directed Purification of Peptides, Waters Application Note. 2004; 720000920EN.

4. cleary R, Lefebvre P. Purification Workflow Management, Waters Application Note. 2006; 720001466EN.

Figure 7. AutoPurify processing of the UPLC analysis of the collect fractions.

Waters, ACQUITY UPLC, UltraPerformance LC, and UPLC are registered trademarks of Waters Corporation. AutoPurification, OBD, Waters Corporation XBridge, AutoPurify, FractionLynx, Application Manager, and The 34 Maple Street Science of What’s Possible are trademarks of Waters Corporation. Milford, MA 01757 U.S.A. All other trademarks are the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720002283EN LB-KP www.waters.com Purification Workflow Management Ronan Cleary and Paul Lefebvre Waters Corporation, Milford, MA, U.S.

INTRODUCTION The software will decide which shallow gradient should be used to perform the purification (Figure 2). A standard requirement for drug discovery screening of synthetic libraries is that the test compounds must have a minimum purity. Purity is based on the area percent of an LC chromatogram from a detector such as UV, evaporative light scattering (ELS), MS with a total ion chromatogram (TIC), or a combination of multiple detectors. If the screening compounds do not meet this standard, purification is required. Managing the flow of samples, subsequent fractions, and all the associated data through this process can often be difficult and time consuming.

This application note illustrates how a sample is efficiently taken through a three-step purification process utilizing the AutoPurify™ Figure 2. TIC chromatogram after purification, with fraction collection indicated by the shaded area. capabilities within the Waters® FractionLynx™ Application Manager for MassLynx™ Software, and the AutoPurification™ System for MS-directed analysis. This comprehensive informatics solution Then, it automatically performs analysis of the collected enables automation from the initial evaluation, through the purifica- fractions (Figure 3). tion, to analysis of the collected fraction.

DISCUSSION

The AutoPurify functionality uses the results of the analytical analysis to determine the purification process. By performing an analytical evaluation of the sample, the presence of the target compound is confirmed and its purity measured (Figure 1).

Figure 3. TIC chromatogram of the analysis of the collected fraction.

Information determined from analysis of the fractions can be used to help with post-purification handling such as fraction pooling and transfer to an evaporator. A report can be exported in different file formats such as .xml, .csv, and .tab, to easily interface with other sample handling software packages.

Figure 1. TIC chromatogram of the analytical-scale analysis of the crude sample. Step 1: Analytical interpretation Step 2: The purification process

In the first of the three-step process, the purity of the target mass is In the second step of the process, purification occurs. The software identified by integrating the chromatogram. In the example shown will determine the purification method best suited to improving the in Figure 4, the area percent of the target determined from the TIC separation by choosing one of six different shallow gradients. Using (22%) is then used to calculate the sample purity. the analytical retention time of the target, the appropriate shallow gradient-based method will be chosen. The area percent can also be determined by total absorbance current, wavelength, or analog signal. The purity of the target is then classi- Shallow gradients, also referred to as narrow gradients, allow fied as “pass,” “tentative,” or “fail,” based on user-defined limits. In for optimal target separation from closely eluting impurities, this example, less than 10% pure means purification will not occur, thus improving the purity of the resulting fraction. Each narrow 10 to 80% purity requires purification, greater than 80% is pure gradient, whose time window is indicated by the colored lines enough, and does not require further purification. (Figure 5), is created to cover a different timed section of the analytical gradient.

Figure 5. Graphical representation of analytical and prep gradients.

Figure 4. Analytical evaluation of mass 357.1 is 22% of the TIC, and the target sample is co-eluting with peak 2. An overlay chromatogram of the two co-eluting peaks, with the spectrum, indicates the potential fraction The analytical gradient is indicated by the dotted black line, and contamination that could occur. shows the solvent change over the course of the gradient to be from 5 to 95% B. With the relationship between the analytical retention time and the elution organic composition known, the software can choose In a manual process, the analyst would evaluate the separation, and which of the narrow gradients will be used to automatically purify the adjust the gradient to achieve the best results. However, in an open samples during the purification stage of the process. access environment or where large numbers of samples are being handled, automation is necessary. When the software evaluates the analytical sample, it creates a browser report defining the recommended strategy. The user has the opportunity to change the strategy if necessary. The part of the report that refers to the strategy is the results pane (Figure 6). In this example, there are several other samples analyzed, but the one that is of interest is that last one on the list, A123008.

34 Figure 6. Browser results pane with sample purity and prep strategy displayed.

The sample in this case eluted at 4.04 min (Figure 7), so the narrow gradient chosen for the purification was “Narrow Gradient Figure 8. Overlay of the chromatograms of the two masses that were C,” the one that targeted the solvent change that occurred between co-eluting earlier, showing the improved separation that was achieved. 4 and 5 minutes. This gradient is denoted by the green line, which Spectra highlight the success also. changes from 24 to 37% organic over 6.5 min, and is defined graphically as below. Step 3: Fraction analysis

With the first two steps of the process complete, the user can also decide to analyze the fractions (Figure 9). AutoPurify creates a sample list containing the fractions required for analysis and automatically runs them.

MS ES+ :358.1 (3) 1.3e+007 100% 357.1 100

% 50

0 Time 2.50 5.00 7.50 10.00 MS ES+ :TIC (3) 1.6e+007 Figure 7. Representation of the narrow prep gradient chosen for the 100% purification of the compound eluting at 4.06 min, with improved 357.1 100 separation showing the isolated peak at 3.74 min collected.

% 50

The improved separation is more clearly displayed when the chro- 0 Time 2.50 5.00 7.50 10.00 matograms of the two co-eluting compounds, as seen in Figure 4, are extracted and their chromatograms reviewed. Figure 8 shows the two Figure 9. Fraction analysis post-collection, and post-fraction mixing by chromatograms of masses 255 and 358, overlaid, and the improved the injector/collector. TIC shows no other compounds present in the collection vessel. separation achieved.

35 To ensure that the portion of the sample taken for analytical Analytical interpretation analysis is representative of the entire collected fraction, it may FractionLynx browsers also include chromatograms and spectral be necessary to pre-mix fractions prior to injection (done with the information that are not shown in this application note. The portion injector/collector). Once homogenized, analysis can be performed of the browser file in Figure 10 shows sample purity and the prep on an analytical scale. strategy decision that was determined after the samples were ana- Automating the process lyzed on an analytical scale.

Automation of the three-step purification process is accomplished The preparative sample list is automatically created and run after the through AutoPurify. analytical analysis. Once the purifications are complete, the results are processed and a new FractionLynx browser report is generated A FractionLynx browser is created after each of the three stages to (Figure 11). display results of the analysis and to report the recommended strategy for the next stage in the process. The software can automatically create and run the list of samples that are to continue to the next step. The user has a choice whether to allow the three stages to run unattended, or to manually review the results of each stage and edit the software’s decision.

The determined strategy can be adjusted as necessary by the user through the interactive browsers that are produced. By automating Figure 11. Purification results, indicating where the fractions were collected, the process, decisions can be made after regular work hours, allowing including fraction volume and spectral purity. Blue = collected fraction of the sample highlighted in the injector plate, green = passed spectral purity the work to continue unattended, saving time and resources. assessment, burgundy = review required, and red = failed purity assessment.

The root name of the data, the sample ID, sample list, and the FractionLynx browser, A123, as shown in Figure 10, are edited by the software and carried through the purification process to make sample Purification process and results tracking easier. Upon completion of the processing of the purification results, a sample list is generated and automatic analysis of the fractions generated is performed (Figure 12).

Figure 10. Browser report created after the analytical evaluation. The resulting strategy is displayed using different colors for the injection plate. Green = mass is found, purity level between 20 and 80%, and sample requires purification; Figure 12. Fraction analysis results, indicating the sample purity of the collected and red = mass is either not found or sample is already pure enough, and fractions. purification will not be performed.

36 Fraction analysis The benefits of using AutoPurify can be measured in time savings, reduced solvent consumption, and overall productivity gains. This is The final report shows the locations of the fractions, chromatograms, noticeable in several main areas: and spectra. The information in the reports can then be easily exported n Automated evaluation of samples before purification prevents in different file formats such as .xml, .csv, and .tab, to easily interface unnecessary purification from being performed by removing with sample handling software packages such as liquid handlers or samples that do not require purification. weighing devices. n Computerized evaluation of samples throughout the entire process saves analysts from having to manually review batches between stages of the process, and enables the subsequent analysis to be performed immediately – without waiting for CONCLUSION the analyst to be present. This application note shows how a library of compounds can easily n Computerized determination of methods required during the and efficiently be purified using the AutoPurify capabilities within process saves analysts from having to make or decide which the FractionLynx Application Manager. The software is capable of gradients should be used to improve separations. automating the entire purification process, from the original analyti- n The use of narrow gradients allows for the use of shorter, cal purity assessment, to purification, and finally to the analysis of more focused gradients, saving time and solvents. the fractions. n Automation from stage to stage allows for unattended operation, combining all the savings of the process. AutoPurify allows the process to be performed intelligently. Analytical results are used to determine if the target is present and its purity. Based on these criteria, only samples that truly require purification continue on through the process. Samples that do not contain the target compound, not enough of the target, or are already pure enough can simply be excluded from purification.

Waters is a registered trademark of Waters Corporation. MassLynx, Waters Corporation AutoPurification, AutoPurify, FractionLynx, and The Science of What’s 34 Maple Street Possible are trademarks of Waters Corporation. All other trademarks Milford, MA 01757 U.S.A. are the property of their respective owners. T: 1 508 478 2000 ©2006-2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001466EN LB-KP www.waters.com MAKING A PURIFICATION SYSTEM MORE RUGGED AND RELIABLE Ronan Cleary, Paul Lefebvre, and Warren Potts Waters Corporation, Milford, MA, U.S.

OVERVIEW

The demand for the number of samples requiring purification continues to grow. This increase requires purifications systems to be able to run more efficiently and with less user intervention. However, there are a number of serious, corporate concerns with running unattended purification. These include losing samples due to system failure, solvent leaks, overflowing waste containers, and solvent reservoirs running dry. Another concern is the verification that the system is actually running properly and collecting fractions as expected.

This application note highlights how the Waters® AutoPurification™ System hardware and software can be utilized to alleviate theses Figure 1. The mass-directed AutoPurification System consists of the 2545 Solvent Manager, 2767 Sample Manager, System Fluidics Organizer, and concerns. Examples include software tools for monitoring solvent PDA detector. usage and that can monitor the number of injections without fraction collection. We also show how the system can be efficiently shut down in case of error to minimize the risk of sample loss. analytical and preparative flow rates. Additional pumps are regularly Finally, we demonstrate how a new splitter can increase recovery added to the system for other purposes, such as post-column split- rates and how a post-fraction collector detector can be used as a ter make-up, at-column dilution (US Patent #6,790,361), off-line quality control monitoring tool. column regeneration, and pre-column modifier solvent addition. Mass spectrometry was added to further increase the selectivity and efficiency of the systems. These components comprise the Waters AutoPurification System. DISCUSSION Solvent monitoring System configuration The various pumps and vessels configured in a purification system System configurations can vary depending on customer applications can be defined in the monitoring software. The volume of solvent and requirements. Waters has developed a purification system based pumped from a solvent reservoir or into a waste container is moni- on input from our customers. tored using the solvent monitor software. The requirement for chemists to be able to make analytical injections Graphical solvent level indicators allow for easy viewing of the sys- to evaluate a sample before purification led to the development of tem status. Each solvent reservoir has information specific to that the Waters 2767 Sample Manager, which has two separate flow paths container, maximum volume, and various warning levels. – one analytical, and one for preparative. A separate and additional flow path allows for fractions to be collected onto the instrument bed The status of the vessels is indicated by symbols, indicating that the for further analysis. This injector/collector requires a solvent deliv- system is either OK, or in Warning or an Acute Warning state. The ery system that is capable of delivering reproducible and accurate response to the warning level is determined by the administrator. A color-coded status page is also available, and can be accessed remotely through the remote status monitor component of the software.

Once all the solvents are defined, monitoring occurs in the background without any user interaction. Any volume of solvent pumped, either during an acquisition or while idle, will be accounted for. Even the amount of solvent used to prime the pump is monitored.

When the software monitoring the solvent vessels identifies a solvent level that has generated a warning condition, multiple notifications and responses can occur, such as: n Warning notification on the instrument page n Color-coded notification on the remote monitoring software n Email condition report sent to primary responsible party Figure 2. Solvent monitoring interface with both graphical and numerical n Terminate the analysis or batch reporting of system status. n Secondary emails can be sent to different individuals, notifying them of the condition of the particular system

Figure 3. Color-coded system status page, with icons that indicate the need to Figure 4. Email configuration with primary and secondary refill or empty the containers. email contacts.

Once the administrator has been notified, they can choose to manage the condition by emptying or refilling the containers as necessary, or allow the software to deal with the error condition and shut the system down safely.

40 Figures 5 and 6. The user can partially add or remove solvents as necessary.

Shutdown software allows the user to configure a response produced when either the warning or acute level is reached: Figure 7. The user can define the number of injections that can occur without fraction collection before the run is ended. n Shut down immediately n Shut down after delay n Shut down after sample n Shut down after batch Additional collectors n Ignore the warning Frequently, analysts find that compounds other than the primary The shutdown procedure configured is linked to a particular shutdown compound of interest are of importance, so it may be necessary to method. This allows for an orderly shut down of the system to occur, capture them in a separate collector. Examples include collection allowing for columns to be flushed and returned to the correct condi- of a starting material or impurities along with the primary target. tions for storage, thus reducing the risk of damage. Another example is collecting all the other major peaks in addition to Tracking failures the primary target. This is shown in Figure 8 with a complex natural product separation. A critical component to ensure rugged and reliable unattended operation is to have the system be able to stop after a defined number of consecutive samples without fraction collection. There are various reasons why a system may not have collected fractions, and yet not be in an error state, such as a blocked splitter or MS sample cone that prevents detection, or a blocked injection port that keeps the sample from being loaded onto the column. User error can also be a contributing factor. Incorrect information such as mass or wavelength can also contribute to fractions not being collected. Figure 8. The top chromatogram shows collection of peaks detected by ELS detection. The lower chromatogram shows the peaks detected by the MS and collected by mass trigger.

41 There is no such thing as a universal detector, so it is possible that some compounds may not be detected. A waste collector can be Figure 10. The upper added to the system, enabling all column eluent not diverted for chromatogram shows the collection earlier to be collected separately. In Figure 8, any of the low-flow split to the fraction trigger detector. The middle sample not collected by either the primary or the secondary collec- chromatogram shows tors was captured in a separate waste collector, thus minimizing the the high-flow split of the sample after using another possibility of any sample loss. commercially available splitter to the waste detector. The lower Splitter performance chromatogram shows the high- flow split of the sample using On any purification system where a destructive detector is being the Waters splitter to the waste used, a splitter is necessary to isolate a portion of the primary detector. flow for analysis, allowing the rest of the sample to be directed to the fraction collector. The flow to the collector must also go through a delay coil to prevent this much faster flow from reaching the collector before the triggering detector has identified the peaks to collect.

Figure 11. Overlay of the trigger and collected fraction trace using a Waters splitter. The collected fraction is the purple trace, and shows little or no peak dispersion.

Figure 9. The Waters splitter is matched to column dimensions for optimized performance. Figure 12. Overlay of the collections with the vertical axis linked. The green trace shows what would have been missed if a non-Waters splitter had been used.

The most important requirement of the splitter is that peak shape and resolution achieved from the column be retained in both the low- and high-flow solvent streams. The low-flow stream is sent to the detectors used to trigger fraction collection. If the peaks’ shapes differ between the triggering detector and the fraction collector, the collection of the fraction will be less than optimal. Laminar flow can cause the peaks on the high-flow side of the system to be larger than the peaks on the low-flow side of the system. This can contribute to decreased recoveries and impure fractions.

We evaluated a new Waters splitter against another commercially available splitter to highlight the improvements that have been made with the splitter technology.

Figure 13. AutoDelay results page with delay time and results export.

42 Collector delay time CONCLUSION

Delay time determination can be easily accomplished with the use of Purification systems should include functionality that allows for AutoDelay software, which will perform injections to determine the unattended operation such as: delay time and confirm injection for the determined delay time. n Solvent monitoring with tiered responses such as email notification Figure 14 shows the effect of delay time on the amount of missed n Solvent monitoring with intelligent shut down fraction detected in the waste detector. The larger the detected n Remote system monitoring peak corresponds to a lower recovery or increased sample loss. n Secondary fraction collection for use with other detectors When the delay time is set optimally there is only a small peak, just n Waste collection to enhance user confidence above the noise. But as the delay time drifts from 1 to 3 seconds away from the optimal, the increase signal becomes more and Flow splitters should not increase band broadening and decrease more substantial. The measured recovery is greater than 99% at fraction recovery rates. The new Waters flow splitters maintain the optimal delay time. With the 3 seconds too early, the recovery equal peak shape for both the high and low flow for optimal fraction is only 60%. recovery and purity.

The AutoPurification System, with technology that allows for rugged and reliable operation, is available from Waters.

Figure 14. Different collection delay values have different responses in the waste detector.

Waters is a registered trademark of Waters Corporation. Waters Corporation AutoPurification and The Science of What’s Possible are 34 Maple Street trademarks of Waters Corporation. All other trademarks Milford, MA 01757 U.S.A. are the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720002285EN LB-KP www.waters.com THE APPLICATION OF MS/MS DIRECTED PURIFICATION TO THE IDENTIFICATION OF DRUG METABOLITES IN BIOLOGICAL FLUIDS Paul Lefebvre, Robert Plumb, Warren Potts, and Ronan Cleary Waters Corporation, Milford, MA, U.S.

INTRODUCTION

The identification of drug metabolites following animal or human volunteer studies is essential to the drug discovery and development and regulatory submissions process. Traditionally, this has been achieved by the use of liquid or gas chromatography coupled to mass spectrometry.1,2 More recently, the use of hyphenated techniques such as LC/NMR and LC/NMR/MS have become more commonplace in the drug metabolism laboratory, allowing a more precise identification of the site of metabolism.3,4

While LC/NMR and LC/NMR/MS are extremely powerful tools, they are typically low throughput and limited in sensitivity. The capacity Figure 1. The Alliance HT System with the Quattro micro Mass Spectrometer. of analytical columns restricts the amount of material that can be loaded on to the column before the column exhibits either volume or mass overloading effects and the chromatographic resolution This application note shows how tandem quadrupole mass spec- is lost. Thus LC/NMR is less attractive for the analysis of highly trometry has been employed for the isolation of the metabolites of potent compounds dosed at low levels or those compounds that common pharmaceuticals from urine. The application also demon- undergo extensive metabolism. In such cases, it is often necessary strates different modes of data acquisition, including scan, MRM, to perform a pre-concentration step, such as SPE or liquid/liquid constant neutral loss, and precursor ion. We also demonstrate how extraction, both of which are time consuming and run the risk of the use of MS/MS-directed purification facilitates the combination of losing of valuable information. samples from several chromatographic runs.

The use of MS-directed purification, using semi-preparative scale columns (typically 19 mm i.d.), is now commonplace within the pharmaceutical industry, especially to support lead candidate METHODS AND DISCUSSION purification. This approach has also been applied to the isolation 5 ® ® of drug metabolites with some success. The extra sensitivity and A Waters Alliance HT System was used with a SunFire™ C18 5 µm selectivity of MS/MS mass spectrometry allows for more precise 4.6 x 100 mm column at 40 °C. Eluent flow was split 1:20 with a selection of drug metabolites. Furthermore, the use of neutral loss Valco tee. 95% of the flow passed the 2996 Photodiode Array (PDA) and precursor ion scanning detection modes facilitates the collec- Detector to the Fraction Collector III. The other 5% of the flow was tion of drug metabolites without the need for prior knowledge of routed directly to the Quattro micro™ Mass Spectrometer equipped compound metabolism. with an ESCi® multi-mode ionization source. Caffeine metabolites methods MS detection

Separation Electrospray negative, 3 kV capillary voltage, 30 V cone voltage, 20 V collision energy. Water/acetonitrile in 0.1% formic acid, 1.25 mL/min total flow gradient. 0 to 5 min: 0%; 5 to 35 min: 0 to 10% B; 35 to Metabolites of interest 35.5 min: 10 to 95% B; 35.5 to 39.5 min: 95% B; 39.5 to 40 min: Figure 3 shows the fragmentation patterns of the ibuprofen 95 to 5% B; 45 minutes end. gluceronide metabolite.7 MS detection

Electrospray positive, 3 kV capillary voltage, 30 V cone voltage, 20 V collision energy (for MS/MS experiments).

Metabolites of interest

Figure 2 shows a portion of the caffeine metabolism pathway by demethylation.6 Target metabolites maintain the methyl group in the 1 position. They also have a common fragment ion, m/z 57.

Paraxanthine, m/z 181 1,7 Dimethylxanthine

Caffeine, m/z 195 1,3,7 Trimethylxanthine Figure 3. Ibuprofen gluceronide metabolite with a common product ion of 1Methylxanthine, m/z 167 m/z 193.

Theophylline, m/z 181 1,3 Dimethylxanthine Common fragment Single quadrupole directed purification of xanthine a methyl in position 1, m/z 57 With single quadrupole directed purification, all ions generated in the source are passed through the quadrupole and detected. This is possible on the Quattro micro Mass Spectrometer by using the scan Figure 2. Metabolism of caffeine by demethylation: metabolites that maintain the methyl group in the 1 position have a common fragment ion, m/z 57. mode of acquisition. Only MS1 is scanned and there is no collision energy or scanning of Q3.

Because all of the ions generated are detected in this mode, Ibuprofen metabolites complex mixtures can contain numerous isobaric interferences. Separation Consequently, multiple fractions can be generated from a single m/z value. Figure 4 shows the collection of the caffeine metabolites Water/acetonitrile/10 mM ammonium formate, 1.25 mL/min total with m/z 167 and 181 detected using only the first quadrupole. flow gradient. 0 to 5 min: 5%; 5 to 35 min: 5 to 60% B; 35 to 35.5 There are eight fractions collected for m/z 167 and five fractions min: 60 to 95% B; 35.5 to 39.5 min: 95% B; 39.5 to 40 min: 95 collected for m/z 181, with additional analysis required to deter- to 5% B; 45 minutes end. mine the fraction of interest.

46 For a peak to be present in the MRM chromatogram, both the specific precursor and the specific product ion need to be detected. For each target, only one fraction was collected.

m/z 181

% OR

0 2.50 5.00 7.50 10.00 12.50 15.0017.50 20.0022.50 25.0027.50 30.00 32.50 35.00 37.50 40.00 42.50 45.00

m/z 167 Figure 4. Fractionation based only on scanning the first quadrupole. %

Tandem quadrupole directed purification: 0 Time 2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50 25.00 27.50 30.00 32.5035.00 37.50 40.0042.50 45.00 MRM collection Figure 6. Fractionation based on MRM acquisition. With multiple reaction monitoring (MRM) data acquisition, MS1 is pre-selected on the precursor mass and MS2 is pre-selected on a specific product ion, as illustrated in Figure 5. Constant neutral loss collection

A second possible mode of fraction triggering is from constant neutral loss acquisition. Here both MS1 and MS2 are scanned in synchronization, as illustrated in Figure 7. When MS1 transmits a specific precursor ion, MS2 looks for a product that is the precursor minus the neutral loss value. If the correct product is present, it MS1 Collision Cell MS2 ststaticatic RF onlyonly stataticic registers at the detector. The constant neutral loss spectrum shows (all masses pass) only the masses of all the precursors that lose the specific mass.

Figure 5. MS/MS MRM data acquisition.

By selectively detecting a product ion, the signal-to-noise ratio is optimized, thus reducing the isobaric interference and allowing only the target to be collected. This mode of acquisition requires MS1 Collision Cell MS2 scascanningnning RF only scscanninganning previous knowledge of the exact precursor and the exact product (all masses pass) ions before purification. Figure 7. MS/MS constant neutral loss data acquisition. Figure 6 shows the MRM acquisition and collection of the caffeine metabolites. The metabolites of interest for isolation have the transi- tions of 181 to 134, and 167 to 110.

47 18 1.0 100 Precursor ion collection m/z 181 % % 18 2.0 A third mode of fraction triggering is from precursor ion acquisition, 0 m/z 100120 140160 180200 0 2.50 5.00 7.50 10.0012.50 15.0017.50 20.00 22.50 25 .0027.50 30.0032.50 35.0037.50 40 .0042.50 45.00 as illustrated in Figure 9. Here, MS1 is scanning and MS2 is fixed

166.9 m/z 167 100 on a specific product ion. If the specific product ion is observed, it is %

% registered at the detector. The spectrum only shows the masses that

114.4 0 m/z 100120 140160 180200 have that specific product. 0 2.50 5.00 7.50 10.0012.50 15.0017.50 20.00 22.50 25 .0027.50 30.0032.50 35.0037.50 40 .0042.50 45.00

Neutral loss of 57 TIC %

0 Time 2.50 5.00 7.50 10.0012.50 15.0017.50 20.00 22.50 25 .0027.50 30.0032.50 35.0037.50 40 .0042.50 45.00

Figure 8. Fractionation based on constant neutral loss acquisition. MS1 Collision Cell MS2 scascanningnning RF onlyonly ststaticatic (all masses pass)

Figure 8 displays the constant neutral loss of 57 acquisition and Figure 9. MS/MS precursor ion data acquisition. collection of the caffeine metabolites with m/z 167 and 181. It shows that two fractions are collected, one for each mass. These fractions contain the target mass and have the specific neutral loss. Fraction collection from a precursor ion acquisition has to be from the TIC, since the precursor mass is unknown. This mode of fraction Applications for fraction collection from constant collection is valuable when the metabolites are unknown, but there is neutral loss acquisition a common fragment of the core compound that can be detected. Mass triggered collection To illustrate the common fragment ion collection capability, Figure With constant neutral loss acquisition, the only peaks detected are the 10 shows the glucuronic acid conjugates collected from the ibupro- ones with the loss of the specific mass, in this case, 57. Depending on fen urine samples using the precursor ion scan mode of m/z 193. the specificity of the loss, numerous ions can be detected. This leads There are three fractions that are collected, m/z 273 (not drug- to complex total ion chromatograms. Therefore, when triggering by related), m/z 397 (hydroxyglucuronide conjugate), and m/z 381 a specific mass, the collected target must contain the precursor of (glucuronide conjugate). interest and have a specific neutral loss.

273.1 Collection triggered on TIC 100 m/z 273 % % 274.7

0 m/z 20 0300 400500 600700 800 When using this mode of acquisition and collection, all the peaks 0 2.50 5.00 7.50 10.00 12.50 15.00 17 .50 20.00 22.50 25.00 27.50 30.00 32.50 35.00 37.50 40.00 42.50 with a specific neutral loss are collected. This functionality is valu- 39 7.0 10 0 m/z 397 % able when the metabolites have a specific loss related to the drug’s % 397.8 398.7 273.1 0 m/z 20 0 300 40 0 500 60 0 0 structure. It could also be used for isolating a class of metabolites 2.50 5.00 7.50 10.00 12.50 15.00 17 .50 20.00 22.50 25.00 27.50 30.00 32.50 35.00 37.50 40.00 42.50

38 1.2 10 0 with a generic loss (e.g., sulfates (–80) or glucuronides (–176)). m/z 381 %

% 382.0 The precursor mass for each fraction can then be extracted and used 38 3.4 0 m/z 20 0 300 400 500 600 0 to aid in the identification of the metabolites. 2.50 5.00 7.50 10.00 12.50 15.00 17 .50 20.00 22.50 25.00 27.50 30.00 32.50 35.00 37.50 40.00 42.50

Parents of 193 TIC

In the constant neutral loss example shown, collection could also %

7 Time have been triggered from the total ion chromatogram (TIC). All peaks 2.50 5.00 7.50 10.00 12.50 15.00 17 .50 20.00 22.50 25.00 27.50 30.00 32.50 35.00 37.50 40.00 42.50 in the –57 TIC would be collected and then additional analysis or data review would be required to find the desired fractions. Figure 10. Fractionation based on the precursor ions of the m/z 193 TIC acquisition.

48 Additional collection options n Constant neutral loss mode can be used for collecting a class of compounds with a target-specific loss or a generic group loss The ESCi multi-mode ionization source enables both ESI +/- and APCI for a broader study, or can be used as a second filter where the +/- acquisition to occur within the same run. This allows for fraction target has to have a specific mass and the neutral loss. collection to be triggered from any of the acquisition channels, thus n Collection in precursor ion mode allows for all the precursors proving useful if the metabolites require different ionization modes. with a specific product ion to be collected, which is valuable Prior to this enabling technology, the only options for collection when the metabolites are unknown, but there is a common would be to split the sample and run in different modes, or rely upon fragment of the core compound that can be detected. time-based fractionation and then analyze all the fractions by both modes to determine the targets. Thus, these different modes of collection add value to a wide variety of applications previously accomplished with more laborious, time The selectivity of the ESCi-enabled fraction collection process can consuming, and less specific methodologies. be further enhanced by the use of mixed triggers. This approach uses Boolean logic strings to trigger collection from multiple data traces (e.g., collection can occur only when Mass A is present and Mass B References is not, or a peak has to be present in two different traces at the same 1. ismail IM and Dear GJ. Xenobiotica. 1999; 29(9): 957-967. time for fractionation). 2. dear GJ, Mallett DN, and Plumb RS. LCGC Europe. 2001; 14(10): 616-624.

3. dear GJ, Plumb RS, Sweatman BC, Parry PS, Robert AD, Lindon JC, Nicholson JK, and Ismail IM. Journal of Chromatography B. 2000; 748: 295-309.

4. dear GJ, Plumb RS, Sweatman BC, Ayrton J, Lindon JC, and Nicholson JK. CONCLUSION Journal of Chromatography B. 2000; 748: 281-293.

Fraction collection with a tandem quadrupole mass spectrometer is 5. plumb RS, Ayrton J, Dear GJ, Sweatman BC, and Ismail IM. Rapid Communications in Mass Spectrometry. 1999; 13(10): 845-854. now possible using four different modes of data acquisition: scan, 6. Bendriss E, Markoglou N, and Wainer IW. Journal of Chromatography B. 2000; MRM, constant neutral loss, and precursor ion, which enables 746: 331-338. improved versatility for triggering options. 7. kearney G, et al. Exact Mass MS/MS of Ibuprofen Metabolites using Hybrid Quadrupole-Orthogonal TOF MS Equipped with a LockSpray Source. Waters n Scan mode has the potential to increase the number of isobaric Application Note. 2003; 720000706EN. inferences detected and collected. n MRM mode is the most selective because it only monitors a specific precursor/product ion transition and greatly reduces the isobaric interferences, but requires previous knowledge of the transition.

Waters, Alliance, and ESCi are registered trademarks of Waters Waters Corporation Corporation. SunFire, Quattro micro, and The Science of What’s 34 Maple Street Possible are trademarks of Waters Corporation. All other trade- Milford, MA 01757 U.S.A. marks are the property of their respective owners. T: 1 508 478 2000 ©2005-2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001129EN LB-KP www.waters.com Evaluating the Tools for Improving Purification Throughput Paul Lefebvre, Warren Potts, Ronan Cleary, and Robert Plumb Waters Corporation, Milford, MA, U.S.

INTRODUCTION

Chemists are constantly looking for ways to improve the overall throughput of their purification system. Time is the limiting factor for throughput, and there are two areas where time savings can be achieved: the amount of time required to perform a separation, and the amount of time between injections. Making the purification system as efficient as possible requires optimizing and minimizing both of these times. The challenge, however, is minimize these times without impacting the purity and recovery of the fractions.

In this application note, we examine tools available for increasing the overall throughput of a purification system. We will use information from the analytical separation to optimize the purification, and will examine the steps required between injections to then determine the most efficient way to minimize run time. Figure 1. The Waters mass-directed AutoPurification System.

OVERVIEW METHODS AND DISCUSSION

In order to correctly compare time-saving techniques, we first Components established a baseline separation to define a standard analysis and The Waters® AutoPurification™ System is comprised of: collection time. We purified 10 drug-like compounds with a generic n 2545 Binary Gradient Module (BGM) 10-minute preparative gradient. This baseline analysis time was then n 2767 Sample Manager used as the comparison time for the analysis performed when the n System Fluidics Organizer (SFO) different time-saving chromatographic functionalities were applied. n 2996 Photodiode Array Detector The major areas for improving throughput are: n 3100 Mass Detector n Decreasing the time required for the analysis n 515 makeup pump n Decreasing the time between injections n Passive flow splitter, 1:1000 n All components are controlled by MassLynx™ One approach for decreasing the analysis time uses shallow or narrow and FractionLynx™ software gradients. Approaches for decreasing the time between injections include column regeneration techniques and automatically ending The 10-sample library consisted of various drug-like compounds the purification run after the desired target has been collected. at a sample concentration of about 20 mg/mL dissolved in DMSO. The chromatographic methods used water with 0.1% formic acid as Gradient Analytical %B %B mobile phase A, and acetonitrile with 0.1% formic acid as mobile Name Retention Time Start End phase B. Methanol was used as the makeup solvent for the prepara- A 0.00 to 1.67 5 20 tive analysis. B 1.67 to 2.84 20 35 Analytical gradient C 2.84 to 4.0 35 50 D 4.00 to 5.17 50 65 SunFire™ C18 4.6 x 50 mm, 5 µm, 1.5 mL/min total flow gradient and E 5.17 to 6.34 65 80 a 10-minute total run time. F 6.34 to 7.5 80 95

Table 2. The narrow gradients used relative to the analytical retention time.

The time window in which the analytical sample eluted defines the conditions for the prep run. For example, if the compound eluted at 4.04 min, then the purification method would ramp up the organic percentage so that is was 50% at 0.5 min.

Figure 2. Analytical gradient table. Baseline throughput

The generic gradient was used to perform the purification of 10 samples and the overall run time was measured. This time is used to Generic preparative compare the improvements.

SunFire C18 19 x 50 mm, 5 µm, 25 mL/min total flow gradient. The same gradient table, as shown in Figure 2, was used. The only difference was the flow rate. Time Retention Run Time Between Sample Narrow or shallow preparative gradient Time (min) (min) Injections (min) SunFire C 19 x 50 mm, 5 µm, 25 mL/min total gradient. The start 18 1 1.18 10 2 and end percent B composition is variable and dependant on the 2 5.20 10 2 sample retention time during its analytical analysis. 3 1.35 10 2 4 4.67 10 2 Time (Minutes) Composition (%B) 5 3.18 10 2 0.00 to 0.5 5 to %B start 6 2.55 10 2 0.50 to 1.67 %B start to %B end 7 2.41 10 2 1.67 to 2 %B end to 95 8 5.06 10 2 2 to 3 95 9 2.02 10 2 3 to 5 End 10 2.63 10 2

Table 1. Narrow gradient table. See Table 2 for percent B start and end. Total Run Time 120 minutes

Table 3. The overall throughput with the generic gradient. The total run time was 120 minutes.

52 Narrow gradients Time Generic Narrow Run Narrow Between Narrow gradients can be used to improve preparative chromatographic Sample Retention Retention Time Gradient Injections resolution.1 However, if the resolution is adequate in the analytical Time (min) Time (min) (min) (min) separation, a shorter narrow or focused gradient can be used to 1 1.18 A 1.38 5 2 increase throughput. The short method will focus its gradient on the 2 5.20 E 1.65 5 2 same organic concentration, but in a shorter time frame. 3 1.35 A 1.74 5 2 4 4.67 D 1.94 5 2 5 3.18 C 1.75 5 2 6 2.55 B 1.90 5 2 7 2.41 B 1.95 5 2 8 5.06 D 2.34 5 2 9 2.02 B 1.30 5 2 10 2.63 B 2.08 5 2 70 minutes = Figure 3. The different narrow gradients possible to focus on either improved Total Run Time resolution or throughput. 1.7 Fold Increased Throughput

Table 4. The overall throughput increases by 1.7 fold when incorporating narrow gradients, compared to using a generic gradient. Figure 4 shows an example of one of the 10 samples being purified by both a generic and a narrow gradient. The target was successfully isolated using narrow gradient D. The results show that the resolution is maintained over the focused section of the gradient Rinsing and equilibration

(the blue bracket). Note that there is a loss in resolution, as expected, It is important for high-quality chromatography that the column is in the non-focused areas of the gradient. This would have to be con- rinsed and re-equilibrated with the appropriate volume of solvent, sidered when the compound elutes at the very beginning or end of typically defined in column volumes. Insufficient rinsing can cause the focused gradient. carryover, and equilibration time also has a significant impact on the overall throughput, with inadequate equilibration leading to retention time variability, poor chromatographic peak shape, or even Generic Gradient Narrow Gradient 5.06 2.34 100 sample breakthrough. The quantity of rinsing solvent is dependant EIC = 270 EIC = 270 upon the sample matrix, the retentiveness of the column, and the elu-

% tropic strength of the rinsing solvent. Typically, two to three column volumes is required to rinse. For equilibration, various articles report 1.99 0 2.00 4.00 6.00 8.00 10.00 1.00 2.00 3.00 4.00 5.00 anywhere from three to 20 column volumes can be used.2-3 0.49 0.88 100 3.20 0.93

2.45 TIC TIC For example, a 19 x 50 mm column has a volume of about 12 mL. 4.76 5.06 % 4.61 2.10 1.60 2.20 1.93 2.34 Two column volumes or 24 mL of 95% B were used to flush the 4.06 6.07 5.58 0.49 2.60 1.56 1.19 3.01 column, and 60 mL of 5% B were used to re-equilibrate the column.

0 Time 2.00 4.00 6.00 8.00 10.00 1.00 2.00 3.00 4.00 5.00 With the gradient flow of 25 mL/min, the flush takes about 1 minute, and the equilibration takes about 2.5 minutes. Figure 4. Comparison of the 10-minute generic and the 5-minute narrow purification. The blue bracket corresponds to the focused area of the gradient, where the resolution is maintained.

53 100 Early termination Area for potential time savings To further reduce the time required for analysis, a software tool can be Chromatographic run time used to automatically end the run after the target has been collected.

Flush and equilibration time The throughput improvements of this feature will be illustrated for

Injection time both generic and narrow gradients.

0.01.0 2.03.0 4.05.0 6.0 7.0 8.09.0 10.0 For either gradient approach used, once the target has finished Time collecting, the gradient will stop and flush with 95% B to wash the Figure 5. Illustration of an injection cycle with chromatographic analysis remaining material off the column. After a defined time of rinsing, time, equilibration and flush time, and injection cycle for next injections time the column will then be re-equilibrated with the initial gradient displayed. The area where time could potentially be saved is noted. solvent. (Note: 2 minutes of equilibration time is performed between injections.)

However, the flow rate can be elevated above optimal chromatographic conditions (30 mL/min for 5 µm packing), so long as the Generic Generic Narrow Narrow system can withstand the overall pressure increase. We found that the Sample Run with Run with flow could be increased to 40 mL/min, only generating an additional Time Regeneration Time Regeneration 1300 psi of backpressure, reducing the flush time to 0.6 min and 1 4.03 3.43 4.23 3.63 the re-equilibration time to 1.5 min, for a 1.5-minute savings. 2 8.05 7.45 4.50 3.90 Off-line regeneration 3 4.20 3.50 4.59 3.99 4 7.52 6.92 4.79 3.19 To increase throughput, a regeneration pump can be used to flush 5 6.03 5.43 4.60 4.00 and re-equilibrate the first column off-line, while the next sample 6 5.40 4.80 4.75 4.15 is running on a second column. 7 5.26 4.66 4.80 4.20 In this method, the run is terminated at 2.5 min for the narrow 8 7.91 7.31 5.19 4.59 gradients, or 7 min for the generic and the next injection started. 9 4.97 4.27 4.15 3.55 The first column is switched off-line and its flush started, while the 10 5.48 4.88 4.93 4.33 second column is put in-line to receive the next sample. As men- Total 58.75 52.75 46.53 40.53 tioned earlier, the time required for the injection to be performed Run min = min = min = min = is 2 min. Time 2.0 Fold 2.3 Fold 2.6 Fold 3.0 Fold Increased Increased Increased Increased The run-time savings for a generic preparative saw a reduction of Throughput Throughput Throughput Throughput 3 min per sample, for a reduction in the total run time from 120 to Table 5. The overall throughput improvement using the run termination function 90 minutes, or a 1.2-fold savings. can range from a two- to three-fold increase, depending on what additional tools are used. Using the regeneration pumps saves 0.6 min per injection when The run-time savings for a narrow gradient was more significant. compared to a single column method. This corresponds to the time required to rinse the column. The re-equilibration time is incorporated into the 2 min to Injection-to-injection time was reduced from 12 min with the make an injection. generic method to 4.5 min using narrow gradients and off-line column regeneration. This reduced the total run time from 120 to 45 minutes, a 2.7-fold savings.

54 Optimized injection routine Two options are available for positioning the sample in the loop. The default setting is to center the sample in the loop, but the sample Throughput can be further improved by reducing the time between centering can be disabled to allow the sample to be more quickly injections. The injection cycle can be divided into three segments: loaded onto the front of the sample loop. n Aspiration of the sample into the needle n Dispensing the sample into the loop When sample centering is removed, it is possible to operate with only n Washing the assembly one wash solvent and to be able to perform this wash at the beginning of the injection sequence, decreasing the injection time. Optimizing the speed of the aspiration enables the sample to be quickly drawn into the needle and holding loop. Care must be taken Cumulative time-savings to ensure the increased syringe speed does not create air bubbles The time required to inject and rinse was reduced from 2 min with in the system. the standard partial loop injection to 0.4 min with the new settings. Once the sample has been drawn into the holding loop, it is dis- Table 6 shows the throughput possible by combining optimized pensed at an optimized flow rate. Care must again be taken to injection settings with the various other tools. ensure that a high-pressure condition does not occur by operating the syringe too quickly.

Original Optimized Default Overall Increase with Tool Total Injection Total Injection Optimized Injection Routine Run Time Run Time Routine Generic 120 104 — 1.2 Generic + 58.75 53.75 2.0 2.2 End Run

Generic + End Run + 52.75 36.75 2.3 3.3-Fold Increased Throughput Regeneration

Narrow 70 54 1.7 2.2 Narrow + 46.53 41.63 2.6 2.9 End Run Narrow + End Run + 40.53 24.53 3.0 4.9-Fold Increased Throughput Regeneration

Table 6. Using optimized injection routines can improve the overall throughput. The improved injection routine has a greater impact when using regeneration because the 2 min for the normal injection is used to re-equilibrate with a single column. But with regeneration, the re- equilibration is done off-line and the injection time is dead time.

55 CONCLUSION References

1. p Lefebvre, A Brailsford, D Brindle, C North, R Cleary, W Potts III, BW Smith, Throughput can be increased by about five-fold using a combina- Waters Poster Presentation, PittCon. 2003. tion of narrow gradients, early run termination, off-line column 2. a.P. Schellinger, P.W. Carr, Journal of Chromatography A. 2006; 1109: 253- regeneration, and an optimized injection routine. This correlates to 266. an 80 percent decrease in run time. 3. UD Neue, American Laboratory. 1997; March. n Narrow gradients can be used to improve throughput, but require additional information about the target. n Off-line column regeneration has a greater impact on throughput as the run time is reduced. n Early run termination improves throughput and reduces the amount of consumed solvent saving both time and money. n Optimizing the wash sequence and adjusting when it is performed will save additional time between injections. n Various combinations of throughput-enhancing tools can be used based on the specific requirements.

Waters is a registered trademark of Waters Corporation. Waters Corporation FractionLynx, MassLynx, SunFire, and The Science of What’s 34 Maple Street Possible are trademarks of Waters Corporation. All other Milford, MA 01757 U.S.A. trademarks are the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001696EN LB-KP www.waters.com A NOVEL APPROACH FOR REDUCING FRACTION DRYDOWN TIME Paul Lefebvre, Ronan Cleary, Warren Potts, and Robert Plumb Waters Corporation, Milford, MA, U.S.

INTRODUCTION

Recent advances in purification technology have shifted the throughput bottleneck from purifying samples to fraction drying. Some of the technologies employed for sample drying include vacuum cen- trifugation, heated nitrogen blow-down, and lyophilization. However, each one has the same rate limiting factor – the quantity of water present. This quantity is dependant on the separation technique used to generate the fractions. The most commonly used technique is reverse phase- (RP-) HPLC, which can generate fractions with the water content as great as 95%.

One approach experimented with is to collect fractions directly onto solid phase extraction (SPE) cartridges. In theory this method is Figure 1. Plumbing diagram for the concentration system. Both fraction collection and concentration was performed on the same mass-directed AutoPurification perfect, but making it automated and rugged has continued to be System. Fraction collection was triggered by MS. a challenge. A drawback to this approach is that a very high flow dilution pump is required to trap the compound on the cartridge. This high flow rate requires a large quantity of sorbent with large volume cartridges, and generates large volume fractions. Another problem METHODS AND DISCUSSION with collection onto SPE cartridges is the possible change in selectiv- The standard components of the Waters® AutoPurification™ System ity that could result in poor trapping or breakthrough of the analyte. were used to perform the fraction concentration. In the plumbing dia-

This application note shows the development and optimization of a gram shown in Figure 1, the aqueous flow out of the gradient pump method that removes the water and reduces the overall volume of the is directed into the first tee (T1). This tee acts as a mixer, diluting the collected fraction. This method works by injecting and trapping the organic concentration of the injected fraction, so that it will not break previously collected fraction onto a preparative column. The fraction through the trapping column. The organic flow out of the gradient is trapped by diluting the loading flow with 100% aqueous mobile pump is directed to a second tee (T2) and is used to elute sample phase. After the trapping has been completed, 100% organic mobile from the column. phase is passed through the column to elute the sample. Collection of Proof of principle the target is triggered by the MS detector and the collected fraction is now in 100% organic mobile. To establish a baseline performance of the method parameters, 10 drug-like compounds were initially purified. These purified fractions were collected in different concentrations of organic solvent and then used as the samples to evaluate the concentration method. The samples were loaded onto a trapping column and eluted in 100% organic solvent. Once it was determined that the initial method was successful, the process was optimized for minimum fraction volume and maximum throughput. The examples shown have initial fraction volumes as great as 30 mL of aqueous/organic and are reduced to as little as 1.5 mL of organic solvent. Purification method Method optimization n 10 mg sample load Once the trapping method was determined to be successful, we looked n Generic 5 to 95% gradient with water/ACN/formic acid into optimizing the conditions. The parameters evaluated included n Fraction volume of 5 to 8 mL with recoveries of greater than 95% the column dimension and packing, the dilution ratio, and the elution flow rate. An initial fraction of 10 mg of diphenhydramine collected in 8 mL of 60% water was the concentration test sample.

Column dimensions

The column must be able to trap the target fraction and yet give a minimum elution volume for the concentrated fraction. Figure 2. Generic 5 to 95% gradient. The maximum flow rate and the minimum loading time were determined to establish a minimum run time. These factors are dependant Concentration method upon the column I.D., particle size, and injection loop. The collected fractions were injected onto the same column as was used for purification. The samples were loaded onto the column with 1.50 98 a loading pump at 6 mL/min 100% A, and 29 mL/min aqueous Purification from a dilution pump. After 6.5 min, the loading and dilution flow is %

0.65 2.77 m/z=235.0 stopped. Now that the sample is retained on the column, the elution -2 Time 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60 is started at 29 mL/min of 100% B. 98

% Acid Concentration 3.04 3.24 3.41 3.53 3.73 3.77 4.10

m/z=235.0m/z=235.0 -2 Time 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50

4.41 98 4.45 RESULTS 4.50

% Base Concentration

4.00 4.04 (50 µL NH4OH) 0.50 Diphen m/z=235.0m/z=235.0 012606_02a 1: Scan ES+ -2 Time 6.91 TIC 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 98 1.58e8

4.40 98 4.48 Acid Concentration %

m/z=167.2 % 1#1,1:1 to 1#1,1:2 Base Concentration -2 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 1: Scan ES+ TIC (200 µL NH OH) 1.58e8 4 012606_01e m/z=235.0 2.74 98 -2 Time 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 2.85 Purification % Figure 4. Two of the remaining three were successful after adding base to

0.660.71 m/z=167.2 fraction. This indicates that these samples should have been purified at a 1#1,1:72 -2 Time 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 basic pH to keep the target neutral.

Figure 3. Seven of the 10 were successfully concentrated in the acidic mobile phase in which they were collected. All recoveries were greater than 85%. 19 x 50 mm trap column

Although the remaining sample was purified using the SunFire™ 5 and 10 µm packing gave the same fraction volume. The only Column, it was not retained on the column during the concentration difference was the system back pressure. process. However, because fraction collection was triggered by MS, no sample was lost. Additional work is required to determine why it was not retained.

58 Diphen Conc 012606_02a 1: Scan ES+ Diphen Conc 10x50 mm 5 µm column 4 mL/min load, 24 mL/min dilution 6.91 TIC 013106_12b 1: Scan ES+ 98 1.58e8 6.70 TIC 98 1.41e8

Elution Flow =

% 24 mL/min

Volume = % 5.8 mL m/z=167.2 1#1,1:1 to 1#1,1:2 -2 Time 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00

Figure 5. Concentration of the test fraction on a 19 x 50 mm column.

m/z=167.2 1#1,1:4 -2 Time 10 x 50 5 µm trap column 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00

The overall flow rate was reduced when the method was transferred to the 10 mm column. By reducing the elution flow rate, from Diphen Conc 10 mm col 4 load / 22 dilution / 12 mL/min elute 020106_14b 1: Scan ES+ 24 to 12 mL/min, the concentrated fraction volume was reduced from 6.79 TIC 98 1.31e8 8 mL to 2.9 mL. Elution Flow = By reducing the flow rate even further, to 8 mL/min, the original 12 mL/min 8 mL of 60% water was reduced to 1.6 mL of 100% organic solvent.

There is minimal loss of the overall speed of the analysis with the Volume =

reduced elution flow rate. The loading and dilution pump operate % 2.9 mL at 24 and 4 mL/min, respectively, until 6.5 min. The flow rate was then reduced to the lower elution flow, accounting for the smaller volume, concentrated fractions.

Improving throughput m/z=167.2 1#1,1:28 -2 Time Sample loading rules 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 n The injection volume must be less than half the volume of Figure 6 and 7. Elution flow was reduced with minimal adjustment to peak width. the sample loop. Because the injection volume was 8 mL, the minimum loop volume was found to be 15 mL. n 3 to 5 times the loop volume is required to clear the sample from the sample loop. The minimum volume found to clear allow it to be trapped onto the column. If the dilution ratio is too all the sample was 45 mL. small, it will cause breakthrough. If it is too large, it will decrease the throughput because of the additional time required to load the Dilution ratio sample. Figure 9 shows the effect of the concentration with varying The dilution ratio (dilution flow/loading flow) is a critical factor dilution ratios. The results show that at a ratio of 4.5 there is a in this method. The dilution ratio is a measure of the amount of jagged breakthrough of the target compound that is not present at aqueous solvent used to dilute the fraction’s organic content to a ratio of 5 or higher.

59 Diphen Co nc 10 mm col 4 load / 22 dilution / 8 mL/min elute Scaling the method 020106_16a 1: Scan ES+ 6.89 TIC 98 7.57e7 Based on the minimum loading time and dilution ratio, it is possible to establish the relationship between the loading time and the total Elution Flow = flow rate (Table 1). 8 mL/min To reduce the loading time to less than 5 min, the table shows that Volume = a loading and dilution flow of 10 and 50 mL/min, respectively, are

% 1.6 mL required. This gives a total flow of 60 mL/min across the column.

Loading Flow Loading Time Dilution Flow Total Flow (mL/min) (minutes) (mL/min) (mL/min) m/z=167.2 5 9.0 25 30 1#1,1:35 6 7.5 30 36 -2 Time 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 7 6.43 35 42

Figure 8. Concentration of the test fraction on a 10 x 50 mm 5 µm column at an 8 5.63 40 48 elution flow rate of 8 mL/min. 9 5.0 45 54 10 4.5 50 60

Diphen Conc 10 mm col 4 load/ 22 dilution / 8 mL/min elute 15 3.0 75 90 020206_14a 1: Scan ES+ 6.94 TIC 98 1.11e8 20 2.25 100 120 Ratio = 5.5 Table 1. Relationship between loading time and total flow. %

Handling increased back pressure

m/z=167.2 n Increase the particle size: a 2x increase equals a quarter of 1#1,1:9 -2 Time 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 the back pressure

Diphen Conc 10 mm col 5 load/ 25 dilution / 8 mL/min elute 020206_19a 1: Scan ES+ 6.96 TIC n Waters SunFire column, 10 x 50 mm, 10 µm 98 1.12e8 n 60 mL/min only generated 2,300 psi Ratio = 5.0

Diphen Conc 10 mm col 5 load/ 25 dilution / 8 mL/min el 021406_09 1: Scan ES+ % 4.86 TIC 98 5.11e7

Concentration of 10 mg test fraction m/z=167.2 1#1,1:16 to under 2 mL in 5.5 min -2 Time with greater than 95% recovery 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00

Diphen Conc 10 mm col 5 load/ 20 dilution / 8 mL/min elute

020206_10a 1: Scan ES+ % 6.88 TIC 98 1.39e8

Ratio = 4.5

m/z=167.2 1#1,1:5 % -2 Time 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.61 7.02

6.47 6.05 6.25 Figure 10. Results from the optimized method. m/z=167.2 1#1,1:40 -2 Time 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00

Figure 9 A-C. Concentration of test fraction with varying dilution ratios.

60 Mass load

One concern with these optimized parameters is the mass load on the smaller trapping column. To evaluate this, the compounds were purified with increasing mass load on the preparative column until overload conditions were achieved. The collected fractions were concentrated using the optimized method. Two examples are shown.

100 Example 1: 60 mg of Ketoprofen 100

Purification: Concentration: The initial purification generated a 10 mL 10 mL fraction 3.6 mL of organic solvent fraction containing about 50% water. The containing with greater than about 50% water 95% recovery %

concentration method successfully reduced % the volume to 3.6 mL of organic solvent with a recovery greater than 95%.

Figure 11 A-B. The purification and concentration of 60 mg of ketoprofen.

100 Example 2: 20 mg Phenyl-tetrazole 100

The purification generated two fractions Purification: Concentration: 10 mL fraction 3.6 mL of organic solvent with a total volume of 18 mL contain- containing with greater than about 50% water 95% recovery ing about 60% water. The concentration % % successfully reduced the volume to 3.2 mL of organic solvent with the recovery greater than 95%.

0 Time 0 Time 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00 7.50 8.00 8.50 9.00 9.50 10.00 0.250.500.751.001.251.501.752.002.252.502.753.003.253.503.754.004.254.504.755.005.255.505.75 When the chromatography begins to overload for the purification on a 19 x 50 Figure 12 A-B. The purification and concentration of 20 mg of phenyl-tetrazole. mm, the fraction will not be completely trapped on the 10 x 50 mm column.

Automatic pooling 100 100 Purification: Concentration: 10 injections of same sample All 10 fractions loaded Fraction pooling on the trapping column 10 3 mL fractions collected on the trap column 30 mL total volume eluted together can also increase throughput. In Figure in 1.5 mL or organic solvent % 13, a 3 mL fraction was collected for each % of the 10 injections. The fractions were then individually loaded onto the trap column and concentrated. A single 1.5 mL 0 Time 0 Time fraction was collected. 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00 7.50 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40

Figure 13. An example of automatic pooling of 10 fraction tubes into a single concentrated fraction.

61 CONSIDERATIONS BENEFITS

The pKa of the target compound should be considered when perform- Dry-down time ing purification. The target compound should be neutral during the Composition Volume Dry Down purification. This means that a basic compound should be run in a Aqueous/Organic 5 to 30 mL 5 to 15 hours basic mobile phase and, conversely, an acidic target should be run in 100% organic 1 to 3 mL less than 30 min an acidic mobile phase. Concentrating the fraction to about 3 mL of organic solvent can be This will result in better loading and chromatography1 and will also accomplished in 6 min. ensure that the collected fraction is not ionized in solution. By being neutral, it is more likely to be successfully trapped during the con- Process enhancements centration process. n Shorter drying times equals more efficient use of the driers. The amount of collected material, in both mass and volume, will n Automatic pooling of multiple fraction tubes reduces the dictate the required system configuration. The volume of the fraction post-purification sample handling. will determine the size loop required. The mass of collected material will determine the column size. Both the loop and column size will determine the overall throughput of the system. Acknowledgements n Ian Sinclair, AstraZeneca In the examples shown, all of the concentrated fractions were trig- n Giovanni Gallo, Waters, Manchester, UK gered by MS. However, this was done only for method development n Paul Rainville, Waters, Milford, MA purposes. It is possible to collect these fractions by UV or even just by time. When collecting by time, each tube has the same volume and organic concentration, so the time required for drying is References constant. With typical fractionation, each tube can have a different 1. Neue UD. Wheat TE. Mazzeo JR. Mazza CB. Cavanaugh JY. Xia F. Diehl DM. volume and organic concentration, so the time required for drying J. Chrom. A. 2004; 1030: 123-134. is variable. This variability can lead to inefficiency, by either drying for too long, or by stopping too early then checking multiple tubes to find that you need to restart for only a few of the tubes.

Waters is a registered trademark of Waters Corporation. Waters Corporation AutoPurification, SunFire, and The Science of What’s Possible 34 Maple Street are trademarks of Waters Corporation. All other trademarks Milford, MA 01757 U.S.A. are the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720002097EN LB-KP www.waters.com profiling PROFILELYNX APPLICATION MANAGER FOR MASSLYNX SOFTWARE Increasing the throughput of physicochemical profiling The client: Physical Chemistry group at a major pharmaceutical company

BACKGROUND

The Physical Chemistry group at a major pharmaceutical company was created to support Discovery projects in hit identification, lead identification, and lead optimization phases with early physicochemical data. The Discovery groups send test requests for selected compounds simultaneously to respective departments via the Chemical Support (CS) team within the Chemistry department. The Chemistry department is where all synthesized compounds are collected and stored. Compounds are sent out for testing according to the requests, as either standard stock solutions or solid samples.

The Physical Chemistry group is made up of three analytical chemists running two LC/UV/MS systems. These systems each consist of a Waters® Alliance® HT System with a 2996 Photodiode Array (PDA) Detector and a ZQ™ Mass Detector, running on MassLynx™ Software. Testing is done in a 96-well plate format.

Among the analyses performed by the team are identification, purity, stability, and solubility tests. ID and purity evaluations are always included in all solubility and stability tests and demand additional processing of data.

THE CHALLENGE

A screen solubility test of 48 samples took approximately 51 hours of analyst time, from when the samples were received to when the data was entered into the database. For a plate containing 48 duplicate samples, the variety of tasks involved: n 4 hours doing sample prep and running the samples n 18 hours in the office collecting compound and plate information – codes, predicted properties, structures – and creating appropriate sample lists n 8 hours evaluating purity n 19 hours doing the solubility calculations n 2 hours inputting the final data into the company’s database.

The analyst would get results well over a week later. The Physical Chemistry group recognized that tests were taking too long to deliver results. They needed to significantly reduce bottlenecks in data management and analysis, as well as instrument resources, to improve their ability to support discovery projects – especially as incoming work volume was increasing. THE SOLUTION As a result of the overall reduction in time, the group is able to analyze more samples, as well as provide the critical information necessary to Creating the proper tools for collecting sample information from make decisions about possible lead candidates more quickly. the database, formatting sample lists, and analyzing the data generated consumed a great deal of analyst time. Because of the success of ProfileLynx with this evaluation, the software will be implemented with other tests within the Physical By implementing ProfileLynx™ – a specialized Application Manager Chemistry group, including solid solubility, stability, and ELogD. for MassLynx that automates processing of physicochemical property analyses – into their existing LC/MS workflow, the chemists reduced the amount of time it takes to perform these tests from 51 to just 20 hours (Figure 1). The office time was reduced from 17 to 2.5 hours. WATERS SOLUTIONS FOR LEAD OPTIMIZATION Because of the improved reporting capabilities of ProfileLynx, the Waters System Solutions for lead optimization provide an auto- solubility evaluation now takes just 4 hours instead of 19. mated, efficient selection process for determining compounds that have potential to become successful therapeutics. These solutions 2003 Lab time Manual Office time combine the strengths of Waters instruments, chemistries, software, Process Evaluation ID/Purity and customer support to assist discovery labs in characterizing 2005 Evaluation ProfileLynx Solubility compounds faster, easier, and more cost-effectively. Database 01020 30 40 50 60 Waters MassLynx software and its ProfileLynx Application Manager Time (hours) streamline data management for physicochemical property Figure 1. Chemist’s time distribution for a screen solubility test of 48 compounds profiling. MassLynx interfaces with upstream data systems to using a manual process (top) vs. ProfileLynx implementation (bottom). build Sample Lists used for data acquisition, while ProfileLynx automates the processing of chromatography-based data for physicochemical property analysis. BUSINESS BENEFIT

While the LC/MS sample analyses was efficient for the screen solubility test, processing the data and interpreting the results required tedious and time-consuming data manipulation and calculation. By introducing ProfileLynx and other tools such as MassLynx templates into their workflow, the customer has saved about 30 hours on the solubility screen for each set of 48 compounds. The time is now used in the implementation of other tests.

Waters and Alliance are registered trademarks of Waters Waters Corporation Corporation. MassLynx, ProfileLynx, ZQ, and The Science of 34 Maple Street What’s Possible are trademarks of Waters Corporation. All Milford, MA 01757 U.S.A. other trademarks are the property of their respective owners. T: 1 508 478 2000 ©2006-2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001793EN LB-KP www.waters.com AN AUTOMATED LC/MS/MS PROTOCOL TO ENHANCE THROUGHPUT OF PHYSICOCHEMICAL PROPERTY PROFILING IN DRUG DISCOVERY Peter Alden, Darcy Shave, Kate Yu, Rob Plumb, and Warren Potts Waters Corporation, Milford, MA, U.S.

INTRODUCTION

The synthesis of large, focused chemical libraries allows pharmaceutical companies to rapidly screen large numbers of compounds against disease targets. Active compounds, or hits, that result from these screens are traditionally ranked based on their activity, binding, and/ or specificity. Turning these hits into leads requires further analysis and optimization of the compounds based upon their physicochemical and ADME characteristics.

The critical factor to consider in physicochemical profiling is throughput. The bottlenecks to throughput include MS method optimization for a large variety of compounds and data management for the large volume of data generated.

Currently, experiments including solubility, chemical and biological stability, water/octanol partitioning, PAMPA, Caco-2, and protein ACQUITY TQD with the TQ Detector. binding are used to generate physicochemical profiles of compounds in drug discovery. The measurement of physicochemical properties from these studies is easily enabled using chromatographic separation and quantitation using LC/MS/MS/UV. While the sample EXPERIMENTAL analyses may be efficient, processing the data and interpreting the results often requires tedious and time-consuming manual manipula- LC conditions tion and calculation. Instrument: Waters® ACQUITY UPLC® System Column: acQUITY UPLC BEH C Column This application note describes an approach to solving these prob- 18 2.1 x 50 mm, 1.7 µm lems by using MassLynx™ Software’s ProfileLynx™ Application Column temp.: 40 °C Manager, a fully automated software package that allows for the Sample temp.: 20 °C design of experiments, data acquisition, and data processing as Injection volume: 5 µL well as report generation. Mobile phase A: 0.1% Formic acid in water To demonstrate the use of this software package, we have devel- Mobile phase A: 0.1% Formic acid in acetonitrile oped an automated UPLC®/MS/MS protocol for data generation. Gradient: Time A% B% Curve Flow The data acquired from multiple assays was processed by a single 0.00 95% 5% 6 0.60 mL/min processing method, all in an automated fashion. As a result, the 1.00 5% 95% 6 0.60 mL/min physicochemical profiling process was significantly simplified and 1.30 0% 100% 1 0.60 mL/min throughput increased. 2.50 95% 5% 11 0.60 mL/min MS conditions Solubility MS system: Waters TQ Detector Software: MassLynx 4.1 with ProfileLynx 2 mM Samples in DMSO ESI Capillary voltage: 3.20 kV Polarity: positive Source temp.: 150 °c Inter-scan delay: 20 ms Desolvation temp.: 450 °C 50 µL 50 µL 50 µL Inter-channel delay: 5 ms

Desolvation gas flow: 900 L/Hr 950950u µLl pHpH 950950u µLl 950950u µLl Dwell: 200 ms 7.4 buffe Buffer r Bubuffer/ACffer/ ACNN ACACNN Cone gas flow: 50 L/Hr

ShakeShak efor fo 24r hours24 ho atur 37s at°CºC 37

Property profiling assays Centrifuge for 15 min at 3000 RPM n A set of 30 commercially available compounds were randomly chosen to demonstrate the ProfileLynx Application Manager. n QuanOptimize™ Application Manager allows for the automated Dilute supernatant 1 to 100 in DMSO optimization of the MS multiple reaction monitoring (MRM) conditions for each compound. AnAnalyzealyz eand an quantitateQud antita againstte standards n Each compound and a reference standard were analyzed by solubility, pH stability, LogP/LogD, and microsomal stability assays based on methods previously published.1,2,3 pH stability n For quantitative experiments, single point or multipoint 200 µM Samples calibration curves were used. in DMSO n To mimic the current practice in discovery labs, 96-well plate formats were used in this study. n pH stability assays were carried out at three different pHs: 50 µL 50 µL stomach (pH 1.0), blood (pH 7.4), and colon (pH 9.4). 50 µL n Solutions were shaken overnight and vacuum filtered through 950 µL 950ul 995050u µLl pH a Sirocco™ plate. 950 µL ph 9.4 0.10.1 MM HC ll 7.4pH 7.4 Bu ffer ammoniumAmmonium buffer Formate n Fractions were quantified against single point 1 µM calibration formate standards. Sample 50 µL at times 0, 5, 19, 15, 30, and 60 min

Neutralize 50 µL Neutralize 50 µL NeutNeutraralizelize 50 50µL samples with samples with sasamplesmple swith wi th 450 µL, 0.02 M sa450mp µLle wates wirth 450 µL, 0.02 M Amammoniummonium HCl HCl Hyhydroxiddroxide e

AnalAnalyzeyze anandQud quantitate anti ta teagainst standards

66 LogP/LogD

Set Alliance HT 50 µL sample + needle depth 475 µL pH 7.4 buffer* Shake to 18 mm to sample 475 µL pH 7.4 octanol** overnight Inject from top phase*** at 37 °C octanol phase 20 µL Manually separate

samples organic and octanol phases Octanol in DMSO into separate vials phase and analyze or ... Aqueous phase 50 µL sample + Shake Set Alliance HT 475 µL water* overnight Inject from needle depth 475 µL octanol** at 37 °C aqueous phase to 0 mm to sample bottom phase*** *Octanol-saturated buffer (or water) **Water-saturated octanol

***Using 2 mL 96-well plate

Microsomal stability

Solution A (4 °C) Solution B (37 °C) Phosphate buffer + Phosphate buffer + NADAPH A + rat liver microsomes NADAPH B

5 µM samples in phosphate Add 100 mL Add 100 mL solution A buffer solution B 50 µL of 5 µM Heat samples 37 °C in 1 mL for 96-well 20 min plate Shake 37 °C Then add Add 50 µL of for 20 min 500 µL acetonitrile 5 µM sample solution + 100 µL of solution A + 500 µL of acetonitrile + 100 µL of solution B

T0 Plate T20 Plate

67 Data processing and report generation n The ProfileLynx results browser contains up to three sections: a results table, the chromatogram, and the calibration curve. n A pass/fail indicator column and user-selected highlight flags allow fast review of the data. n The chromatogram is interactive for manual integration if needed.

Solubility browser LogP/LogD browser

Metabolic stability browser pH stability browser

68 DISCUSSION CONCLUSION n The 30 compounds were analyzed with the LC/MS/MS protocol Using the ProfileLynx and QuanOptimize Application Managers including MS MRM parameter optimization, MS acquisition allows for: method creation, data acquisition, data processing, and n Automated MS method development and data acquisition. report generation. n A single approach for data processing and report generation n The data generated from the variety of assays were all from multiple assays. processed with the same software automatically. n Complete and automated analysis, processing, and reporting. n A single report was created for the 30 compounds that n Increased laboratory throughput. contained results from all property profiling assays, increasing throughput. n Results are displayed in an interactive, graphical summary References

format based on sample or experiment. 1. kerns E. Journal of Pharmaceutical Sciences. 2001; 90 (11): 1838-1858. n Additional improvements to throughput were achieved for 2. US Pharmacopia. 2000; 24: 2236.

the LogP/LogD assay by utilizing the needle height adjustment 3. di L, Kerns E, Hong Y, Kleintop T, McConnell O. Journal of Biomolecular of the Alliance HT system to inject directly from the two phases Screening. 2003; 8(X). of the octanol/water mixture without the need to manually separate the two phases.

Other assays supported: n Protein binding (plate or column) n Membrane permeability (PAMPA, Caco-2, etc.) n Chromatographic hydrophobicity index (CHI) n Immobilized artificial membrane

Waters, Alliance, ACQUITY, ACQUITY UPLC, and UPLC are registered trademarks of Waters Corporation. MassLynx, ProfileLynx, Waters Corporation QuanOptimize, Quattro micro, Sirocco, SunFire, and The Science 34 Maple Street of What’s Possible are trademarks of Waters Corporation. All other Milford, MA 01757 U.S.A. trademarks are the property of their respective owners. T: 1 508 478 2000 ©2005-2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001239EN LB-KP www.waters.com OPTIMIZATION Synthetic Reaction Monitoring Using UPLC/MS Marian Twohig, Darcy Shave, Paul Lefebvre, and Rob Plumb Waters Corporation, Milford, MA, U.S.

INTRODUCTION

Once a chemical hit is found through a library screening process and is verified, optimization of the compounds’ desired properties takes place. This step involves an iterative process of synthesis and reactivity measurement of the new compounds to further develop drug candidates into the lead phase.

Because these reactions may take a long time, chemists need to know as soon as possible if their syntheses are proceeding as desired. This means utilizing measurement capabilities that require minimal sample preparation and provide a fast response giving low detection limits. Another advantageous property might be the ability to measure multiple parameters simultaneously.1

High throughput approaches can provide important time Figure 1. The ACQUITY SQD for synthetic reaction monitoring. savings in the optimization of process parameters. Open access LC/MS is replacing TLC as a reaction monitoring tool.2 Sample preparation of reaction mixtures can be as minimal as filtering and dilution before injecting into the LC/MS system. This allows fast turnaround of results to allow the chemist to advance to the EXPERIMENTAL next step. Chromatographic separations were carried out using an ACQUITY ® ® The purpose of this application note is to demonstrate the UPLC System coupled to an ACQUITY SQ Mass Detector. PDA and advantages of speed and ease of use that self-service UPLC® with ELS signals were collected simultaneously. Samples were analyzed photodiode array (PDA)/evaporative light scattering (ELS)/MS using gradients less than 1 minute. For chromatographic flexibility, detection brings to reaction monitoring studies. a column selection module was added. LC conditions

LC system: Waters® ACQUITY UPLC System

Column: acQUITY UPLC BEH C8 Column 2.1 x 30 mm, 1.7 µm Column temp.: 45 °C Flow rate: 800 µL/min Mobile phase A: 0.1% Formic acid in water Mobile phase B: 0.1% Formic acid in acetonitrile Gradient: 5 to 95% B/0.7 min MS conditions To illustrate the functionality of such a system, the synthesis of atenolol (Figure 2) was used as a reaction model. The increase MS system: Waters SQ Detector in the formation of atenolol was monitored, as was the decrease Ionization mode: eSI positive/ESI negative in the intermediate 4-hydroxyacetamide3 (Figure 3). A reaction Capillary voltage: 3.0 KV by-product 4-hydroxyphenylacetic acid was also observed. Cone voltage: 20 V Source temp.: 150 °C Desolvation temp.: 450 °C O O Desolvation gas: L/Hr

Cone gas: 50 L/Hr NH2 OH Acquisition range: 100 to 1300 m/z Scan speed: 10,000 amu/sec Note: A low volume micro-tee was used to split the flow to the ELS and SQ. OH OH ELS conditions 4---4Hydroxyphenlyacetamide -Hydroxyphenlyacetic Acid

C8H9NO2 C8H9O3 Gain: 500 N2 gas pressure: 50 psi Drift tube temp.: 50 psi Sampling rate: 20 points/sec Atenolol, C H N O PDA conditions 14 22 2 3 Figure 2. Structures of atenolol and 4-hydroxyacetamide. Range: 210 to 400 nm Sampling rate: 20 points/sec

Increase Decrease

RESULTS AND DISCUSSION Atenolol 4-Hydroxyacetamide t=5 min During the compound optimization stage of a discovery cycle, medicinal chemists are not only interested in determining the key t=45 min structural features responsible for activity and selectivity, but also what structural changes need to be made to improve these character- t=50 min istics. Because the reactions necessary to bring about these changes may take many steps, chemists need to be sure they are progressing as expected during the course of the reaction synthesis. t=60 min Time

Figure 3. UPLC/MS chromatograms. The reaction mixture was sampled at various time points.

72 The ACQUITY SQD is capable of scan speeds of up to 10,000 intermediate 4-hydroxyphenylacetamide ionizes in both positive amu/sec. Consequently, it is possible to employ a large number and negative ion mode (Rt 0.34 min) and 4-hydroxyphenylacetic of scan functions in a single run while still maintaining adequate acid (Rt 0.39 min) only ionizes in negative ion mode. peak characterization. The fast scan speed is essential for this The OpenLynx™ Open Access Application Manager, part of functionality, as peak widths of 1 second or less are common with MassLynx™ Software, allows chemists to walk up to a terminal the use of UPLC. Scanning multiple functions allows confirmation and log in samples while entering the minimum information of compound synthesis to be obtained on reaction components required to run the samples. whether they ionize in positive ion mode or negative ion mode, ESI or APCI. The total cycle time of the method was 1 minute 20 The OpenLynx OALogin screen shown in Figure 5 allows the seconds, facilitating increased sample throughput. administrator to set up the system such that the user only needs to input the information requested, and then upon completion, A single run can also provide UV spectral information and an select the Login Samples button. This will either tell the user the estimation of compound purity at low wavelengths. ELS detection designated autosampler position, or confirm the position that the is based upon the degree to which solute particles scatter light. user has chosen, and ask for confirmation of position before it It has been known to give rise to similar responses for related will run the sample. In addition to a simplified sample submission compounds.4 The signal can give a tentative estimation on the process, the OpenLynx Application Manager can then process relative quantities of the components present (Figure 4). It is also data automatically and produce a summary report that can be an alternative detector to UV, which depends on the presence of emailed or printed as desired. a chromaphore. As can be seen from Figure 4, atenolol ionizes in ESI positive ion mode (retention time 0.28 min). The reaction The information contained in the summary report is viewed via the OpenLynx browser shown in Figure 6. It clearly defines what components are found or not found. Chromatograms and spectra

0.34 are generated based on the processing parameters set up by the 5.0e-2 0.39 PDA administrator in the OpenLynx method.

AU 2.5e-2 0.28 0.49 0.0 0.10 0.20 0.30 0.40 0.50 0.60 0.70

0.35 50.000 ELS U

LS 25.000 0.29

0.000 0.10 0.20 0.30 0.40 0.50 0.60 0.70

0.28 100 ESI Positive % 0.34 0 0.10 0.20 0.30 0.40 0.50 0.60 0.70

100 0.39 0.35 ESI Negative %

0 Time 0.10 0.20 0.30 0.40 0.50 0.60 0.70

Figure 4. UPLC/PDA/ELS/MS chromatograms. Figure 5. The OpenLynx single page login.

73 The described system and software combination can autonomously evaluate large numbers of samples, with a cycle time of 1 minute 20 seconds. Data can then be automatically processed and a summary report can be generated. The scan speed capabilities of the ACQUITY SQD make it possible to better characterize narrow chromatographic peaks. This has become a necessity since the advent of sub-2 µm particle technology, where chromatographic peaks can be 1 second wide or less. The fast scan speed allows the chemist to extract as much data as possible per injection by switching between APCI and ESI as well as positive and negative ion modes.

Open Access gives the chemist a walk-up system that is flexible for analytical data acquisition. It runs as a complete system, from sample introduction to end results. Figure 6. The OpenLynx Application Manager browser. The use of the fast-scanning MS along with the throughput of UPLC technology allows the chemist to obtain high quality and comprehensive data about their compounds in the shortest possible time. CONCLUSION This combined with intelligent open access software allows informed decisions to be made faster, thus supporting the needs of the modern During the compound optimization stage of a discovery cycle, drug discovery process. medicinal chemists are not only interested in determining the key structural features responsible for activity and selectivity, but also what structural changes need to be made to improve these char- References acteristics. Because the reactions necessary to bring about these changes may take a long time, chemists need to be sure they are 1. analysis and Purification Methods in Combinatorial Chemistry, Wiley-Interscience. (5): 87-123. progressing as expected. 2. LC/MS Applications in Drug Development, Wiley-Interscience. 96-106.

By using a walk-up UPLC/MS system, chemists were able to quickly 3. a Synthesis of Atenolol using a Nitrile Hydration Catalyst. Organic Process Research and Development. 1998; 2: 274-276. and easily monitor their reactions, noting the relative amounts of starting materials and products. They were also able to note the 4. kibbey, C.E. Mol. Diversity. 1995; I: 247-258. formation of any side products and make necessary alterations to their reaction protocol to minimize these.

Waters, ACQUITY, ACQUITY UPLC, and UPLC are registered trademarks Waters Corporation of Waters Corporation. MassLynx, OpenLynx, and The Science of What’s 34 Maple Street Possible are trademarks of Waters Corporation. All other trademarks are Milford, MA 01757 U.S.A. the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720002258EN LB-KP www.waters.com THE WATERS ACQUITY UPLC SYSTEM Time and cost savings in an Open Access environment The client: A multi-national research-based pharmaceutical corporation

BACKGROUND

A laboratory supporting the medicinal chemistry department of a large global pharmaceutical firm relied on HPLC/MS systems in an open access environment to provide 150 synthetic chemists with critical information about the success of their reactions. The synthetic chemists wanted to ascertain quickly what compounds their reactions have made and whether any of the molecules are known to be toxic.

To get the information they need, the medicinal chemists literally walk up to one of 21 open access systems configured for the purpose, add their sample to the cue, select one of three pre-set HPLC/MS scouting methods and walk away. Minutes later the results are emailed back to them.

At this facility, each open access system handles 600 to 700 samples per month; in 2004 the lab ran a total of 204,000 samples, with much higher numbers expected for subsequent years.

The average run time for an HPLC/MS scouting method is 6.6 minutes. Turnaround time in this overall, the wait for results high-throughput environment is critical. As the lab manager has said, “Anything I can do to save any has been cut in half, while amount of time, I do it.” solvent consumption has been cut by 85 percent. CHALLENGE

The demands placed on the medicinal chemistry department for high-quality new drug candidates dictate that speed is of utmost importance. Despite this lab’s best efforts to reduce turnaround times by pushing their HPLC methods to the limits, the wait for results was sometimes as long as an hour. Things needed to change in order to reduce drug development timelines and cost.

Despite the larger workload, the lab manager had set as his goal a five-minute – or less– turnaround time for results. This ambitious goal clearly required a new approach.

Another key concern for this laboratory manager: injection reproducibility. When chemists are tracking a reaction, any shift in retention times from one analysis to the other is a red flag and suggests that something unintentional might have been created in the reactor.

THE SOLUTION

In 2004, the laboratory acquired a Waters® Open Access ACQUITY UltraPerformance LC® (UPLC®) System, which they put on the front-end of a single quadrupole Waters ZQ™ Mass Spectrometer. The goal was to see by how much they could shorten the run time of their scouting methods–without losing sensitivity or resolution. By eliminating as little as one minute per analysis, the lab could save 80 percent savings by comparison. With no increase in lab space, 3400 hours of total analysis time and increase the number of tests and further savings captured in consumables and solvents, the lab they perform by 34,000. now has a strengthened investment strategy for increasing capacity and productivity going forward.

BUSINESS BENEFIT

The support laboratory began to see their work pay off with UPLC in ways they hadn’t imagined. In short order, they have reduced what WATERS AND UPLC was a 6.6-minute run to just 2.3 minutes, a three-fold improvement The Waters ACQUITY UPLC System synergistically combines instru- in overall run time. Now, sub-two-second peak widths are standard mentation, column chemistries, software for data acquisition and and the lab manager has reported, “I can offer my clients the same processing, and support services, creating a singular solution with peak capacity in one-half the time.” superior sensitivity, resolution, efficiency, and sample throughput. Overall, the wait for results has been cut in half, while solvent When coupled with Waters MS Technologies, UPLC provides a level of consumption has been cut by 85 percent. separation, quantification, and characterization previously unattain- Moreover, the lab manager has reported getting more than 2500 able with traditional HPLC. injections on a single column without any degradation in results. UPLC today is employed by companies to bring their laboratories “I am extremely impressed with the robustness of the column – measurable improvements in analytical sensitivity, resolution, and very happy,” he has said. speed. Ultimately, these firms are looking for meaningful ways to Perhaps the most important benefit of the Open Access ACQUITY increase laboratory productivity, decrease operational costs, facili- UPLC® System relates to the increase in the number of samples tate product development, and increase revenue generation. expected in the near future. The lab manager anticipated an increase of 15 to 20 percent in the next year, which would normally require WATERS OPEN ACCESS SOLUTIONS the addition of up to four complete LC/MS systems, at a cost of over Waters Open Access systems give chemists the ability to analyze their $600,000 in capital investment. Add to that increases in much-needed own samples close to the point of production by simply walking up to laboratory space, service, and maintenance and consumables. the LC/MS system, logging their samples, placing their samples in the The lab manager has been able to develop an alternative plan to system as instructed, and walking away. As soon as the analysis is achieve the same increase in sample capacity by replacing the inlets completed, sample results are emailed or printed as desired. System on two of their existing systems with ACQUITY UPLC Systems. configuration and setup is enabled through a system administrator who This could be achieved for $120,000 in capital investment, an determines login access, method selection, and report generation.

Waters, ACQUITY UPLC, ACQUITY UltraPerformance LC, and UPLC Waters Corporation are registered trademarks of Waters Corporation. ZQ and The 34 Maple Street Science of What’s Possible are trademarks of Waters Corporation. Milford, MA 01757 U.S.A. All other trademarks are the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 June 2007 720001371EN LB-KP www.waters.com

Waters, ACQUITY, ACQUITY UPLC, ACQUITY UltraPerformance LC, Alliance, ESCi, UltraPerformance LC, and UPLC are registered trademarks of Waters Corporation. ApexTrack, AutoPurification, AutoPurify, FractionLynx, i-FIT, LCT Premier, LockSpray, MassLynx, ODB, OpenLynx, ProfileLynx, Quattro micro, Waters Corporation Quattro Premier, QuanLynx, QuanOptimize, Sirocco, SunFire, XBridge, ZQ 34 Maple Street and The Science of What’s Possible are trademarks of Waters Corporation. Milford, MA 01757 U.S.A. All other trademarks are the property of their respective owners. T: 1 508 478 2000 ©2007 Waters Corporation. Printed in the U.S.A. F: 1 508 872 1990 August 2007 720002320EN LB-KP www.waters.com