A highly automated 5 pump, 4 detector super-critical fluid chromatography (SFC) mass spectrometry (MS) system for purification in drug discovery A new system that is a work in progress, but we have initial data to show Qing Ping Han and Mark J. Hayward* Lundbeck Research USA 215 College Rd. Paramus, NJ 07652
Outline Rationale and progress Why SFC/MS Business objectives Instrument design Instrument features Separation examples Chiral Achiral Summary of the experience thus far
We bought our system to meet chiral needs: Why use chiral separations? In drug discovery, many compounds are chiral In some cases, individual enantiomers have unique and valuable biological properties In vitro and in vivo biological tests need to be performed using individual enantiomers, so that the stereo specific biological properties of drug candidates can be measured Often, chiral resolution is most efficiently achieved using column chromatography Reviews on the medical benefits of chirality in pharmaceuticals: I. Agranat, H. Caner, J. Caldwell, Nat. Rev. Drug Discov., 1, 2002, 753. H. Caner, E. Groner, L. Levy, I. Agranat, Drug Discov. Technol., 9(3), 2004, 105.
Why use super-critical fluid chromatography mass spectrometry (SFC/MS) system? SFC can be used with modifier (normally MeOH) gradient to perform generic gradient separation of wide range of compounds (including polar ones) in a way that is generally complementary with RP-LC. SFC is generally the best way to perform generic gradient elution for normal phase separations (sufficiently broad composition range). Most of the best chiral columns work best in normal phase. SFC is faster and has better loading than NP-LC. (of which we do a lot) SFC is MS compatible. This is a crucial efficiency component because it permits collection of only the particular desired peaks. Use of CO 2 as solvent is less expensive than liquid solvents. Fractions are conveniently (and efficiently) obtained in a minimal volume of modifier for easy recovery of purified compounds, with no water in fractions minimizing evaporation bottleneck and saving energy. In our hands, chiral method development takes the least number of injections when using gradient SFC to screen column and mobile phase conditions. Early efforts in preparative SFC/MS with open bed collection: T. Wang, M. Barber, I. Hardt, D.B. Kassel, Rapid Comm.Mass Spectrom., 15, 2001, 2067. X. Zhang, M.H. Towle, C.E. Felice, J.H. Flament, W.K. Goetzinger, J. Combin. Chem., 8, 2006, 705.
Business objectives for our gradient SFC/MS based chiral purification platform Routinely inject 100 mg and yield 10 g per day (1g/hr) of high purity (95-99% ee) for any number of chiral compounds (1-20) with >99% success rate (crucial for achieving trust, throughput and fast turnaround) Where desired, achiral purification of 100 compound parallel synthesis libraries (generally, when there is a need that can t be readily solved with LC/MS based purification) Flow 100g/min Open bed collection under atmospheric conditions to simplify process and enhance safety (must be safe and efficient) Ease of operation, use same equivalent LC/MS software to reduce labor on operation and learning curves (MassLynx / FractionLynx) Informatics: main stream software platform to ensure minimal workload on data handling and integration for efficient process (NuGenesis SDMS, crucial for transparency and trust)
Much of our SFC design philosophy comes from our established approach toward RP-LC/MS based purification RP-LC/MS Purification System Schematic All components under full software control (MassLynx V4.1) [except automatic heaters] Back-flush regenerating pump Back-flush solvent selections (ACN, 5% acetic acid and DMF) MS Photo diode array ELSD heater Makeup pump splitter Fraction collector 6-pos. column selectors choose up to 6 column chemistries Injection port heater SunFire C18 XBridge C18 Inertsil C8 Inertsil C18 Column water bath Dilution solvent selections (ACN, 50/50 ACN water mixture, etc..) At column dilution pump Waste level sensor and auto switcher Waste UV heater mixer Concentrated buffers at 1-4 M: NH 4 COOH, NH 4 COOCH 3, CH 3 COOH, NH 3, H 2 Oetc Concentrated buffer pump Waste barrels B A Binary pump Degassers Acetonitrile MilliQ Gradient water purification and autodelivery system 1
Planned SFC/MS Purification System Schematic All components under full software control (MassLynx V4.1) Back flush/regeneration P50 CO 2 + P50 Modifier (awaiting software) heater AD OD 515 515 AS Splitter OJ Photo diode Array-2996 heater 6-pos. column selectors choose up to 6 column chemistries Dilution solvent selections (alcohols, 50/50 alcohols and CO 2 mixture, etc..) MS-ZQ ELSD-2420 Fraction collector Injection port heater At column dilution pump (Thar analytical FDM) (awaiting software) Waste level sensor and auto switcher Waste UV (2487) SIII Make up pump G L S heater mixer PR 40 PSI CO 2 vent Diethylamine, triethylamine, isopropylamine, ammonium formate, formic acid etc in alcohols Concentrated buffer pump (515) (awaiting software) P-200 CO 2 P-200 modifier G700 with Bulk Tank Waste barrels Degassers Alcohols All components implemented except P50(x2), FDM, and third 515 pump
Instrument photo
Instrument photo
CO 2 Source Photos House CO 2 we do nothing but use it
Key components of the SFC/MS purification system Gas-Liquid Separator (GLS) The flow stream passes back pressure regulator at this point Depressurization of CO 2 occurs at this step The significantly increased volume of gaseous CO 2 needs to be dealt with, rather than letting it flow into collection tubes, which will cause all sorts of problems. GLS: Minimal CO 2 at collection tip Most gaseous CO 2 is vented from the top of GLS Aerosols usually accompanied with this process is mostly eliminated and/or effectively controlled. This ensures minimal sample loss and enhanced safety GLS: smooth flow into collector The residual liquid flow is guided through the bottom of the chamber to open-bed fraction collector. Separation pump adjusts with separation gradient to maintain steady liquid flow through the GLS. Works surprisingly well. Open-Bed collection Collecting in same tubes/racks used in LC to ensure safety and eliminate complexity for otherwise required customized collection design, workload and cost is lowered significantly, and system robustness is enhanced by ease of operation (collecting into EPA tubes - 27 x 100 mm)
Effect of separation temperature on carbamazpine peaks carbamazepine-4 5.0 30 o C 2.25 40 o C 50 o C 2: Diode Array 230 Range: 5.515 4.0 3.0 2.0 1.0 Time 2.10 2.15 2.20 2.25 2.30 2.35 2.40 2.45 2.50 Peak shape doesn t change much with increasing temperature compared with RP HPLC separations. Also, the effects can be counter-intuitive (RT increasing with increasing T opposite expected). However, temperature can still be very helpful with selectivity (see next).
Temperature tuning the separation Temperature can dramatically enhance selectivity, sometimes in unexpected ways A B C D Mixture of endo/exo confirmations and R/S isomers at (A) 30, (B) 40, (C) 50, and (D) 60 C
Mobile phase without buffering Effect on peak shape as a function of loading 50 mg impramine-6 2.75e-1 2.53 100 mg 2: Diode Array 320 Range: 2.192e-1 2.5e-1 2.25e-1 25 mg 2.0e-1 1.75e-1 1.5e-1 Imipramine 1.25e-1 1.0e-1 7.5e-2 5.0e-2 2.5e-2 Time 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 when loading increases, peak shape suffers
Mobile Phase Buffering Controlling separation under high mass loading 100 m g, 5% to 20% M EOH+0.2% DEA, 100 G/MIN, bp120, sp 300,40oC im pram ine-8 2.91 2.75e-1 2.5e-1 2.25e-1 2.0e-1 No additive 0.2% DEA Imipramine 2: Diode Array 320 Range: 3.736e-1 A U 1.75e-1 1.5e-1 1.25e-1 1.0e-1 7.5e-2 5.0e-2 2.5e-2 Time 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 Buffering can help a lot with peak shape under high loading conditions SFC peak shape becomes much better with adding 0.2%DEA in MeOH. We believe adding a 515 pump & 6-way valve will add significant value by allowing method selection of 6 buffer choices and their concentrations in the separation approach proven in RP-LC/MS based systems
Added Automation Features: Waste UV chromatogram Of course, useful in setting collection delay parameters Waste UV chromatogram to show compound was collected (not lost). This helps eliminate time consuming discussions about where did all my compound go? I know I had xxx mg! Convincing nature of data presentation minimizes need for post purification QC for ordinary compounds (95% purity threshold cmpds, i.e. libraries). Sample Report: Sample 195 Vial 5,1:1 ID 22100-090-001 File 4 mixture-t5 Date 07-Jul-2009 Time 16:17:55 Description QPH 2: UV Detector: TIC 4.0e+2 2.0e+2 1: MS ES+ :237+238+259 % 100 75 50 25 0 WasteUV_2487 % 100 75 50 1.85 2.29 a 4.23 3.36 0 1.00 2.00 3.00 4.00 5.00 6.00 2.27 a 0 1.00 2.00 3.00 4.00 5.00 6.00 25-0.28 0 1.84 a 3.35 4.24 6.36 5.84 6. 5.32 0 1.00 2.00 3.00 4.00 5.00 6.00
Added Automation Features: ELSD collected mass estimation Sample Report: ELSD Characteristics 2: UV Detector: TIC 6.0e+2 4.0e+2 1.90 2.55 6.064e+2 Range: 6.064e+2 Sample 244 Vial 5,1:1 ID 22100-089-016 File an-noc Date 13-Jul-2009 Time 13:58:29 Description analytical Mass based detection (not concentration) Fairly analyte independent +/- 20% accuracy readily achievable Automated inclusion in FractionLynx report 2.0e+2 Time 0 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 1: MS ES+ :189+190+211 % ELSD_2420 % 100 75 50 25 100 1.90 1.3e+008 0 Time 0 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 75 50 1.91 2.57 5.7e+004 25 0.81 0.90 0.44 1.26 1.55 2.22 3.15 3.39 3.66 3.94 0 Time 0 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 Time Amount [ mg ] 1.91 24 2.57 15 We still need to test extent that variable split is affecting calibration and accuracy (If so, we could switch to the usual 10000/1 split with 100 fold dilution)
Various separation examples: A. Chiral B. Achiral
Chiral
Example 1: Flurbiprofen on AD column (well known chiral example, loading test for instrument installation) 100 mg, 5 to 30 % MEOH, 100 G/MIN, bp120 sp 280,40oC flurbiprofen-t7 2.71 3.36 4.0 2: Diode Array 250 Range: 5.515 2.0 flurbiprofen-t6 4.0 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 2: Diode Array 2.76 3.46 250 Range: 5.512 2.0 flurbiprofen-t5 4.0 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 2: Diode Array 1.97 2.57 250 Range: 5.341 2.0 Time 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 Top and middle: 100 mg and 50 mg/injection with 5 to 40% MeOH in 5 min gradient; bottom: 50mg/injection and 15% MeOH with isocratic 100g/min and 30x150 mm AD-H column, BP 120 bar
Example 2: Separation and purification of an enantiomeric mixture by prep SFC/MS (in-house compound) 10 to 40% IPA, AD 2x25,80g/m in, bp120,sp258 AF19670-15 Sm (M n, 3x2) 2.34 2: Diode Array 254 Range: 7.097e-1 5.0e-1 2.5e-1 3.69 20% IPA 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 AF19670-16 Sm (M n, 3x2) 2: Diode Array 3.35 254 6.0e-1 Range: 6.994e-1 4.29 4.0e-1 2.0e-1 10 to 40% IPA in 5 min Tim e 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00 A D, 2 0 % IP A, 3 5 o C, 1 2 0 b p AF19670-PK1b 2.65 2: Diode Array 250 Range: 2.344e-1 1.5e-1 1.0e-1 Analytical analysis of fraction 1 and 2 5.0e-2 3.14 5.14 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 AF19670-PK2b 4.51 2: Diode Array 250 Range: 1.422e-1 1.0e-1 5.0e-2 1.86 7.83 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Tim e Peak shape is better using gradient. ee is >99% for fraction 1 and 2 Condition: prep SFC AD, 20x250, 80g/min, bp120 bar Analytical SFC AD-H, 4.6x250, 4g/min, bp120 bar, 20%IPA
Example 3: Purification comparison of enantiomeric mixture by prep SFC/MS and NP HPLC (in house compound) AE94944- t5 S m (M n, 3 x 2) 1.75e-1 1.5e-1 1.25e-1 SFC, 10 to 40% MeOH, 80 g/min Loading: 20 mg 3.61 4.77 3 : Dio d e A rra y 254 Range: 2.048e-1 1.0e-1 7.5e-2 5.0e-2 AD-H 2x25 cm 5 µm In both cases 2.5e-2 Time 1.00 2.00 3.00 4.00 5.00 6.00 7.00 NP HPLC 5% IPA 95% Hexane 14ml/min Loading: 8 mg Under SFC conditions, baseline separation in 7 min. with single tube per peak, but NP-HPLC takes 30 min to separate enantiomers partially and requires collection in many tubes.
Example 4: Purification comparison of enantiomeric mixture on OJ and AS columns using prep SFC/MS (in-house compound) AE88128-t16 4.0e-1 OJ 20x250 4.18 5.63 2: Diode Array 254 Range: 4.074e-1 2.0e-1 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 AE88128-t10 Sm (Mn, 3x2) 2: Diode Array 3.43 254 Range: 1.455 1.0 5.0e-1 0.96 AS 20x250 1.21 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Time 8.00 4.32 OJ gives better separation under same SFC conditions SFC condition: 10 to 40% MeOH in 6 min, 80g/min, BP 120 bar 75 mg loading and single tube per peak collection
Example 5: Purification comparison of enantiomeric mixture by prep SFC/MS and NP HPLC (in house compound) NP-HPLC: cycle time 23min 40 mg/injection IA column - 2x25 cm, 5 um UV detection must fish out relevant tubes Resulting ee is 90% SFC/MS: cycle time 5 min 60 mg/injection. AD-H column - 3x15 cm, 5um UV & MS detection one peak desired, one tube collection Resulting ee is 100% 2.00 1.80 1.60 1.40 1.20 A U 1.00 0.80 0.60 0.40 0.20 0 2.00 4.00 6.00 8.00 10 12.00 14.00 16.00 18.00 20 Comparison: SFC is 7 fold more productive, far less laborious, and delivers higher quality!
Achiral SFC is the go-to technique for highly polar compounds particularly where the molecular differences are near the polar group(s)
Example 1: Separation covering diversity of samples, neutral, basic and acidic (well known standards easy stuff part of installation tests) 2.01 3.48 254 Range: 2.044 1.8 1.6 1.4 2.50 4.34 1.2 1.0 8.0e-1 6.0e-1 4.0e-1 2.0e-1 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 Antipyrine, Carbamazepine, ketoprofen and sulfamethazine Conditions: flow 100 g/min, 5 40% MeOH @ 10%/min, 100 bar, ethylpyridine column, 30 x 150 mm 1 ml injection (100 mg, 25 mg of each)
Example 2: Separation of challenging basic drugs: desimipramine, imipramine and trimipramine (gives RP-LC/MS based approaches a real run for their money!) desipramine_impramine-trimi[ramine-2 2.39 2: Diode Array 230 Range: 5.454e-1 4.5e-1 4.0e-1 3.5e-1 3.0e-1 2.09 2.5e-1 2.0e-1 3.24 1.5e-1 1.0e-1 5.0e-2 Time 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 Peak 1 Trimipramine, peak 2 Imipramine, and peak 3 Desimipramine Conditions: flow 100 g/min, 5 40% MeOH @ 10%/min, 100 bar, ethylpyridine Column, 30 x 150 mm, 1 ml injection (100 mg ea)
Example 3: Purification comparison of reaction mixture by prep LC/MS and SFC/MS (in house compound) Prep RP-LC/MS can separate desired product and starting material under acidic conditions, but product peak is broad and split. m/z = 366 Prep SFC/MS can readily separate starting material (m/z = 258) and desired product (m/z = 366). Product peak is clean without nearby interferences. m/z = 366 m/z = 258 m/z = 258 TIC TIC While LC/MS works, SFC/MS looks like the way to go
Example 4: Purification comparison of reaction mixture by prep LC/MS and SFC/MS (in house compound) Prep LC/MS: desired product and starting material are partially co-eluted m/z = 364 Prep SFC/MS fully separates desired product (m/z = 364) from starting material (m/z = 288) m/z = 364 m/z = 288 m/z = 288 We couldn t get LC/MS to work. SFC/MS was straight forward.
Example 5: Purification of achiral product isomers by prep SFC/MS with chiral column (in house compound) AD-H, 3x15cm, 30:70 IPA/CO 2, 100g/min, 280 nm A mix of isomers (meta/para 55:44) 15:44:12 separated by SFC 250 04-Mar-2010 mg loading AF28962-1c Sm (Mn, 3x4) 2: Diode Array Complete co-elution with RP HPLC 230 Range: 5.078e-1 4.5e-1 4.0e-1 Time 3.24 6.49 Height 503336 204468 Area 72270.71 58490.75 Area% 55.27 44.73 3.5e-1 3.0e-1 2.5e-1 2.0e-1 1.5e-1 1.0e-1 5.0e-2 Time 0 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10 We couldn t get LC/MS to work. SFC/MS was straight forward.
Summary Our new Waters/Thar gradient SFC/MS based chiral purification platform has been setup for only short time and several promised components (pumps) have not yet been implemented. Nevertheless, preliminary results show that SFC100-MD-1 system works well for chiral and achiral separation and purification and is much better than NP HPLC in terms of efficiency (speed and loading). In many regards, our initial impression is that the system works better than we expected. 100 mg/injection has been achieved for both chiral and achiral purification. Analytical data of post purification fractions show that ee% is >99% for all enantiomers. We also have successfully implemented additional ELSD and UV detectors into the system. We believe that with addition components, on column dilution system with CO 2 and modifier, gradient back-flush system, separate buffer delivery sub-system, we will meet all our initial goals: routinely inject 100 mg and yield 10 g per day of high purity (95-99% ee) for any number of compounds (1-20) with >>99% success rate. With this new hardware, we will continue to experimentally push the envelope on purification production and performance. Of course, as progress is made, we will continue to report it.