Anal. Chem. 2000, 72, 4629-4633
Mixed-Mode Anion-Cation Exchange/Hydrophilic Interaction Liquid Chromatography-Electrospray Mass Spectrometry as an Alternative to Reversed Phase for Small Molecule Drug Discovery Mark A. Strege,* Stephanie Stevenson, and Steve M. Lawrence
Lilly Research Laboratories, A Division of Eli Lilly and Company, Lilly Corporate Center, Indianapolis, Indiana 46285
Within pharmaceutical drug discovery, significant needs currently exist for the analysis and purification of structurally diverse samples prior to or immediately following high-throughput screening. These processes are required to facilitate rapid and accurate biological profiling, structural determination, and resupply of new drug candidates. Reversed-phase high-performance liquid chromatography (RP-HPLC) coupled with electrospray ionization mass spectrometry (ESI-MS) for both analytical and preparative applications has become the small molecule separation/ detection tool of choice for meeting many of these needs. However, the separation selectivity provided by RP-HPLC has been limited to the hydrophobicity-based resolution of relatively nonpolar sample components, and for highthroughput drug discovery applications, no sufficient alternative procedures have been identified. In this investigation, a mixed-mode anion-cation exchange/hydrophilic interaction chromatography (ACE-HILIC) method has been developed to provide both direct compatibility with ESI-MS and evaporative light-scattering detection (ELSD) and separation selectivity highly orthogonal to RP-HPLC. The technique employed silica-based smallpore weak ion exchange resins eluted with a combined aqueous and pH gradient. A diverse set of dipeptide probes was employed for the elucidation of the relative contributions of three retention mechanisms. ACE-HILICESI-MS-ELSD should prove useful for the analysis and purification of compounds from both biological (e.g., natural products) and synthetic (e.g., combinatorial chemistry) sources of molecular diversity. High-throughput screening (HTS) represents one of the most important technologies currently in place within the pharmaceutical industry for the identification of biologically active molecules as potential drug candidates. The success of an HTS operation can be presumed to be directly dependent upon the molecular diversity of the samples present within the compound libraries screened. These libraries have generally consisted of traditionally synthesized compounds, combinatorial chemistry compounds, and natural products. Regardless of the specific origins of the molec* To whom correspondence should be addressed: (phone) (317) 276-9116; (fax) (317) 276-5281; (e-mail)
[email protected]. 10.1021/ac000338b CCC: $19.00 Published on Web 08/24/2000
© 2000 American Chemical Society
ular diversity, organic compounds typically originate in solution as mixtures that require the application of separation technology for purification. For the facilitation of rapid and accurate biological profiling, structural determination, and resupply of biologically active samples identified by HTS, significant emphasis has recently been directed toward the analysis, purification, and characterization of components prior to screening.1 Reversed-phase high-performance liquid chromatography (RPHPLC) coupled with electrospray ionization mass spectrometry (ESI-MS) for both analytical and preparative applications has become the separation/detection tool of choice for meeting many of these needs.2-4 RP-HPLC has provided rapid, high-resolution separations of samples in mixtures, while ESI-MS has facilitated on-line characterization via molecular weight determination and structural fragmentation. However, the separation selectivity provided by RP-HPLC has been limited to the hydrophobicitybased resolution of relatively nonpolar sample components (polar compounds present a significant challenge for RP-HPLC), and for small molecule drug discovery applications, no sufficient alternative procedures have been identified. Additionally, a procedure common within the pharmaceutical industry has been the fractionation of samples mixtures by single or multiple preparative RP-HPLC cycles followed by analysis of the fractions via RPHPLC-ESI-MS, i.e., the same separation technique repeated twice. This sequence of processes could tremendously benefit by the incorporation of a sample resolution scheme based upon a set of complementary separations utilizing mechanisms based upon different physical properties of the sample components (i.e., “orthogonal” separations). Recently a technique known as hydrophilic interaction chromatography (HILIC) was adapted for the analysis of samples by HPLC-ESI-MS for drug discovery applications.5,6 Through the establishment of an enriched water layer on the surface of the polar stationary phase into which sample components could partition, this method successfully provided the separation and characterization of polar compounds that could not be analyzed by RP-HPLC. HILIC was therefore based upon a mechanism (1) Greig, M. Am. Lab. 1999, 31 (24), 28-32. (2) Zeng, L.; Kassell, D. B. Anal. Chem. 1998, 70 (20), 4380-4388. (3) Zeng, L.; Burton, L.; Yung, K.; Shushan, B.; Kassell, D. B. J. Chromatogr., A 1998, 794 (1-2), 3-13. (4) Strege, M. A. J. Chromatogr., B 1999, 725, 67-78. (5) Strege, M. A. Anal. Chem. 1998, 70 (13), 2439-2445. (6) Strege, M. A. Am. Pharm. Rev. 1999, 2 (3), 53-58.
Analytical Chemistry, Vol. 72, No. 19, October 1, 2000 4629
opposite to that of RP-HPLC but still operating within the hydrophobicity-hydrophilicity continuum. The power of mixed-mode cation exchange chromatography/ hydrophilic interaction chromatography was first recognized by Zhu et al., who utilized a large-pore (300 Å) strong cation exchange packing and mobile phases of high organic content with a sodium perchloride gradient for the separation of mixtures of peptides 10 amino acids in size or larger.7,8 These researchers found that for large molecules the technique rivaled or even exceeded the performance of RP-HPLC in several cases. The use of a nonvolatile salt for elution rendered the procedure incompatible with ESI-MS, however. Additionally, the separation of small molecules by ion exchange/hydrophilic interaction chromatography was not demonstrated. In this investigation, through the novel integration of both anion and cation exchange technology with HILIC, a mixed-mode anion-cation exchange/hydrophilic interaction chromatography (ACE-HILIC) method is reported which provides both resolution highly orthogonal to RP-HPLC for small molecules and direct compatibility with ESI-MS and evaporative light-scattering detection (ELSD). The chromatography technique utilized silica-based, small-pore weak ion exchange resins eluted with a combined organic and pH gradient for sample resolution. A diverse set of dipeptide standards was employed for the elucidation of the relative contributions of the three retention mechanisms of ACEHILIC. EXPERIMENTAL SECTION Apparatus. The HPLC system employed for this investigation was an Alliance model 2690 separations module (Waters Corp., Milford, MA). ESI spectra were collected on a Platform LCZ mass spectrometer (Micromass Ltd., Cheshire, U.K.) scanning 1602000 m/z every 2 s with positive/negative ion switching. Capillary and sample cone potentials were set at 3000 and 50 V, respectively. A Sedex model 55 ELSD (Richard Scientific, Novato, CA) was employed using a drift tube temperature of 40 °C. Reagents. Ristocetin A, the peptide standards, glucose, lactose, glacial acetic acid, and ammonium acetate were purchased from Sigma Chemical Co. (St. Louis, MO), and vancomycin, tylosin, cephalosporin C, toyocamicin, pimericin, streptovaricin C, and semipurified fermentation broth material were provided by Eli Lilly and Co. (Indianapolis, IN). All standards were prepared in 50% ethanol at a concentration of 1.0 mg/mL with an injection volume of 10 µL, unless otherwise indicated. The chromatographic packings used for this study were the TSK-Gel DEAE-2SW, and CM2SW (250 × 0.46 mm, 125-Å pores, 5-µm particles) purchased from TosoHaas (Montgomeryville, PA), PolyWAX and PolyCATA (200 × 0.46 mm, 200 and 300 Å, 5 µm) from The Nest Group (Southboro, MA), and PEI and CBX (150 × 0.46 mm, 300 Å, 5 µm) from Metachem Technologies (Torrance, CA). HPLC-grade acetonitrile (ACN) was obtained from Burdick & Jackson (Muskegon, MI). Methods. Ammonium acetate (10% w/v) buffer was adjusted to pH 5.5 via the addition of glacial acetic acid. Mobile phase “A” (0.05% ammonium acetate, pH 5.5, 90% ACN) was prepared by the addition of a buffer stock solution and the required volume of water to a volumetric flask, followed by the addition of ACN to a (7) Zhu, B.-Y.; Mant, C. T.; Hodges, R. S. J. Chromatogr. 1991, 548, 13-24. (8) Zhu, B. Y.; Mant, C. T.; Hodges, R. S. J. Chromatogr. 1992, 594, 75-86.
4630
Analytical Chemistry, Vol. 72, No. 19, October 1, 2000
level several milliliters below the mark. After mixing and degassing in an appropriate vessel, ACN was added to the mark. The 5% (v/v) acetic acid mobile phase “B” was prepared in a similar manner but without the use of a stock solution. The chromatography columns were connected in series and eluted with a flow rate of 1.5 mL/min and a 96-min 0-100% B gradient with a 30min reequilibration between injections. All analyses were performed at ambient temperature. Unless otherwise indicated, RP-HPLC conditions employed in this study consisted of a Symmetry C18 packing (150 × 4.6 mm, 5 µm, Waters Corp.) eluted with a 48-min 2-90% ACN gradient in the presence of 0.05% ammonium acetate, pH 5.5, buffer. For all analyses, a 150 µL/min flow was diverted from the HPLC column effluent into both the ESI source and the ELSD. All chromatographic separations were performed at ambient temperature, and the analyses were performed in replicate to ensure reproducibility. Data collection and processing were performed on a PC using Masslynx 3.1 software (Micromass Ltd.). RESULTS AND DISCUSSION ACE-HILIC-ESI-MS Method Description and Performance. The ACE-HILIC-ESI-MS method was designed to combine the separation capabilities of ion exchange and HILIC for the analysis of sample components of unknown structure, such as compounds originating from natural product extracts. Ion exchange has generally been acknowledged as offering the greatest degree of orthogonality to RP-HPLC among known chromatographic methods, particularly for protein and peptide separations,9 by providing a retention mechanism based upon electrostatic interaction. HILIC represents a technique previously demonstrated to provide retention of both charged and noncharged polar species.5,6,10 Using the ACE-HILIC method and the TSK-Gel packings, sample retention occurred as the system was equilibrated for loading in 6.5 mM ammonium acetate, pH 5.5, 90% ACN via three mechanisms: (i) the adsorption of acidic components to the cationic sites within the diethylaminoethyl (DEAE)-2SW packing, (ii) the adsorption of basic components to the anionic sites within the carboxymethyl (CM)-2SW packing, and (iii) the hydrophilic interaction of both charged and noncharged components with both stationary phases under the HILIC conditions provided by the presence of high organic. Sample component elution was facilitated as a gradient was developed to 100% B (5% acetic acid, 50% ACN) through (i) the protonation of acidic compounds at pH ∼3, resulting in desorption from the anion exchange packing, (ii) the protonation of the anionic sites within the weak cation exchange resulting in desorption (this approach has been demonstrated for the purification of peptides),11 and (iii) the removal of the HILIC effect via the decrease in organic concentration. While the use of a weak cation exchanger is required for ACE-HILIC, the anion exchanger may be either weak or strong. For these preliminary studies, the combination of the two 250-mm columns in series resulted in a relatively large total column volume (∼8.3 mL), and a long gradient (96 min), equilibration time (30 min), and high flow rate (1.5 mL/min) were employed to facilitate effective chromatography. (9) Cunico, R. L.; Gooding, K. M.; Wehr, T. Basic HPLC and CE of Biomolecules; Bay Bioanalytical Laboratory: Richmond, CA, 1998; Chapter 9. (10) Alpert, A. J. J. Chromatogr. 1990, 499, 177-196. (11) The Nest Group, Inc. Technical Applications Note P17 4.29.98, 1998.
Figure 1. ACE-HILIC separations of (A) tylosin, (B) vancomycin, and (C) cephalosporin C represented by ELSD chromatograms obtained using TSK-Gel DEAE-2SW and CM-2SW packings and a 96-min gradient from 0.05% ammonium acetate, pH 5.5, 90% ACN to 5% acetic acid, 50% ACN. The “x” indicates a system peak corresponding to a contaminant present in the glacial acetic acid employed in the mobile phase B used for the latter two separations.
Figure 2. Separations of a semipurified microbial fermentation extract sample obtained using (A) RP-HPLC-ESI-MS and (B) ACEHILIC-ESI-MS as represented by total negative ion mass spectrometry signal chromatograms. Nine individual sample components were tracked via their unique ESI-MS spectra as indicated by the dashed lines for the demonstration of separation orthogonality.
Table 1. Relative Retention of Compounds in ACE-HILIC vs RP-HPLCa retention time (min)
elution order
compound
RP-HPLC
ACE-HILIC
RP-HPLC
ACE-HILIC
streptovaricin C cephalosporin C vancomycin ristocetin A toyocamcin tylosin pimericin
3.2 4.2 7.8 11.6 12.4 22.8 32.3
4.6 91.0 68.8 79.7 9.3 21.5 31.5
1 2 3 4 5 6 7
1 7 5 6 2 3 4
aThe conditions used to generate these data are described in the Experimental Section.
The chromatography generated by ACE-HILIC was of a quality comparable to that typically generated by RP-HPLC, as was demonstrated by the ELSD chromatograms of three natural product compounds displayed in Figure 1. Together as a set, tylosin, vancomycin, and cephalosporin C represented a range of structural diversity (hydrophobicity, ionic character, MW ranging from 400 to 1500, etc.) that would be expected to challenge most chromatographic systems. The peak marked by the “x” represented a contaminant present in one of the lots of glacial acetic acid employed for two of these separations. The retention times and order of elution of a larger set of test standards are presented in Table 1, and the data provided evidence for the orthogonality of ACE-HILIC in comparison to RP-HPLC. Additionally, strong ESI-MS and ELSD signals were obtained for each of the test compounds, demonstrating the compatibility of these detection techniques with ACE-HILIC. Further evidence demonstrating that ACE-HILIC was complementary to RP-HPLC was generated through the analysis of a semipurified sample of a microbial fermentation extract containing approximately 50-100 unknown components. The mixture was analyzed by both methods (in this case, a 48-min 25-40% ACN gradient was employed for RP-HPLC), and the total negative ion signal ESI-MS chromatograms are displayed in Figure 2. The unique ESI-MS spectra corresponding to a set of nine components eluting within the RP-HPLC chromatogram in Figure 2A were
Figure 3. Separations of a semipurified microbial fermentation extract sample obtained using (A) RP-HPLC-ESI-MS and (B) ACEHILIC-ESI-MS as represented by total negative ion mass spectrometry signal chromatograms. Four individual sample components were tracked via their unique ESI-MS spectra as indicated by the dashed lines for the demonstration of the enhancement of resolution by ACEHILIC.
extracted and then used to search the ACE-HILIC separation in Figure 2B for the corresponding retention times in the latter elution profile. The dashed lines crossing paths between the two chromatograms indicated significant separation orthogonality. A second evaluation of these data is diplayed in Figure 3., where a subset of four closely eluting components within the RP-HPLC system revealed significantly higher resolution within the ACE-HILIC separation. Investigation of ACE-HILIC Retention Mechanisms. To gain an understanding of the nature and magnitude of the three separation mechanisms that comprise ACE-HILIC, a series of experiments was performed using a set of dipeptides and tripeptides as test probes. Within this set of standards, Arg-Leu and Arg-Lys were chosen to represent moderately and strongly basic small molecules, respectively, Glu-Gly and Glu-Glu moderately and strongly acidic compounds, and Gly-Ala-Tyr a neutral molecule. The first set of ACE-HILIC experiments served to investigate the effects of “decoupling” the ion exchange interactions from the HILIC contributions to determine the relative impacts of the retention mechanisms. With the exception of the ACE-HILIC Analytical Chemistry, Vol. 72, No. 19, October 1, 2000
4631
Figure 4. Retention time (normalized as described in the text) plotted vs separation method for the peptide standards (9) Arg-Lys, (×) Arg-Leu, (b) Gly-Ala-Tyr, (O) Glu-Gly, and (0) Glu-Glu as obtained using ion exchange chromatography in the anion exchange packing (AE), the cation exchange packing (CE), and both packings (ACE) in comparison to retention by mixed-mode chromatography (ACE-HILIC). In the three ion exchange experiments without HILIC, the packings were eluted using a 96-min gradient from 0.05% ammonium acetate, pH 5.5, 30% ACN to 5% acetic acid, 30% ACN. The data representing the acidic, neutral, and basic standards are linked by solid, long dashed, and short dashed lines, respectively. Arg-Lys was not eluted by ACE-HILIC under these conditions; the data point representing this standard was placed at a retention time corresponding to the maximum number of column volumes cycled through the packings under the parameters employed for this experiment.
retention control data, the information plotted in Figure 4 was obtained via ion exchange chromatography in the absence of HILIC. To remove the HILIC mechanism, the concentration of ACN in both mobile phases was reduced to 30% (HILIC retention has previously been reported to become negligible at ACN concentrations of less than 50%).10 The separate contributions of the anion and cation exchange packings are demonstrated in Figure 4 by the data sets labeled “AE” and “CE”, respectively. Under these conditions where only a single column was employed, the gradient run time was reduced to 48 min. For single column analyses, the corresponding analyte retention plotted in Figure 4 was normalized with respect to column volume per unit time to facilitate direct comparison with the data obtained using two columns in series. Under the conditions utilized for ACE-HILIC chromatography, the highly basic test compound Arg-Lys was not eluted. Elution of Arg-Lys could be achieved, however, simply by increasing the number of column volumes of mobile phase B employed for elution from the TSK-Gel packings. Only the acidic test compounds were retained by the anion exchanger alone, while the retention of all of the compounds by the cation exchange packing demonstrated the influence of the acidic environment of the chromatography upon establishing a significant positive charge within each of the test compounds. Both packings connected together in series provided the retention data points displayed labeled as “ACE” in Figure 4. In comparison to the ACE data, the retention of all of the test standards was observed to increase by 50-75% when the mixed-mode ACE-HILIC separation was employed. These results clearly demonstrated the significant contribution of the HILIC effect to the mixed-mode separation. The second set of experiments was designed to investigate the relative contributions of the anion and cation exchange packings to the mixed-mode separations occurring in the presence of HILIC. The five test standards were analyzed using the anion 4632 Analytical Chemistry, Vol. 72, No. 19, October 1, 2000
Figure 5. Retention time (normalized as described in the text) plotted vs separation method for the peptide standards (9) Arg-Lys, (×) Arg-Leu, (b) Gly-Ala-Tyr, (O) Glu-Gly, and (0) Glu-Glu as obtained using mixed-mode chromatography obtained through the use of the anion exchange packing (AE-HILIC), the cation exchange packing (CE-HILIC), and both packings combined in series (ACE-HILIC). The data representing the acidic, neutral, and basic standards are linked by solid, long dashed, and short dashed lines, respectively. Arg-Lys was not eluted by ACE-HILIC or CE-HILIC under these conditions; the data points representing this standard were placed at retention times corresponding to the maximum number of column volumes cycled through the packings under the parameters employed for these experiments.
exchange packing (AE-HILIC), the cation exchange packing (CE-HILIC), and both packings together in series (ACE-HILIC) (see Figure 5). The data obtained using the single columns was again normalized as described earlier to facilitate direct retention comparison. The elution of Arg-Lys was also not achieved under the chromatography conditions employed for both ACE-HILIC and CE-HILIC. The results of Figure 5 indicated that, in AE-HILIC, all retention of both the basic and neutral analytes was lost. However, in CE-HILIC, the retention of the acidic dipeptides decreased but was still partially preserved, while the basic and neutral compounds chromatographed approximately the same in the presence of both columns. These results suggested that, under the acidic conditions employed for ACE-HILIC, the anion exchange packing contributed electrostatic and HILIC retention of acidic compounds only. In this environment (pH
