High Throughput Analysis of Chiral Compounds Using Capillary Electrochromatography (CEC) and CEC-Mass Spectrometry with Cellulose Based Stationary Phases

Abstract

To fulfill the ever growing demand for rapid chiral analysis, this research presents an approach for highthroughput enantiomeric separations and sensitive detection of model chiral analytes using capillary electrochromatography (CEC) with UV and MS detection. This was achieved utilizing a short 7 cm CEC columns packed with cellulose tris (3,5-dimethyl-phenylcarbamate) (CDMPC) or sulfonated cellulose tris (3,5-dimethylphenylcarbamate) (CDMPC-SO₃) chiral stationary phases (CSPs) applying outlet side injections in CEC-UV. The separation performance was compared between CDMPC and CDMPC-SO₃ CSPs for rapid enantio-separation in CEC-UV mode. In addition, using a high sensitivity UV-flow cell in combination with outlet side injections, the S/N and hence the limit of detection of chiral drug could be improved. The 7-cm packed column was also used with traditional inlet injections for CEC coupled to a low-cost single-quadrupole MS. While outlet side injection was not possible in CEC-MS due to instrumentation constraints, the combined use of a short 7 cm column packed with CDMPC-SO₃ CSP provided several fold higher throughput. Both CEC-UV and CEC-MS with short packed bed has the potential for a simple, sensitive and cost-effective method for enantiomeric drug profiling in biological samples.

1. Introduction

Capillary electrochromatography (CEC) with packed columns has demonstrated great potential as an alternative technique to HPLC for analytical-scale separations because of its excellent separation efficiency, high enantioresolution, better sample loading capacity and low consumption of expensive and exotic chiral stationary phase (CSP) (1). However, the number of publications on chiral CEC separation is much smaller compared to chiral HPLC. The main reason for this phenomenon may be due to its technological hurdles, i.e., requirements for somewhat higher operator skill to handle relatively fragile packed CEC columns. Another reason is lack of dedicated CEC instrumentation for column conditioning, capability of performing gradient separations, as well as lack of potential interest among commercial investors due to low financial returns. On the other hand, when compared to the use of moving chiral selector added to the run buffer in traditional CE, the use of commercially available CSP for HPLC if properly packed in the capillary column could generate stable EOF for high efficiency CEC separations. Therefore, CEC can be coupled to both UV and mass spectrometric (MS) detection. The use of packed column CEC-MS if cleverly designed has the added advantages of low operation costs and high S/N. A review on chiral separations using packed column CEC and CEC-MS showed a total of 63 papers published since 1999 (1). A wide variety of CSPs used in HPLC have been applied in developing packed and monolithic CEC (2–11).

Based on their excellent efficiency, broad chiral selectivity, and stability, polysaccharide CSPs have been immensely successful and broadly studied for CEC (12–27). However, a major limitation of these polysaccharide CSPs i.e., cellulose tris (3,5-dimethylphenylcarbamate) (CDMPC) for CEC is very long analysis time (28). This is because the use of a neutral polysaccharide coating on the silica particles hinders the EOF primarily by blocking the accessibility of ionizable groups (e.g., amino groups on aminopropyl silica or silanols on bare silica) to the counterions in mobile phase. Many groups have developed different techniques and new CSPs to counteract this aforementioned drawback of longer elution time on polysaccharide stationary phases in CEC. The group of Birod et al. has decreased the loading of the CSP from 20% (w/w) to 2–5% (w/w) allowing more exposure of the ionizable groups. Unfortunately, this approach reduced the chiral resolution of the columns (15). Zou’s group has worked with positively charged polysaccharide CSPs covalently bonded with cellulose phenylcarbamate derivatives. The authors used this CSP with a short 8 cm column and outlet side injection to increased sample throughput (25,27). Because the secondary amino groups on the silica gel surface acquire positive charge with acidic mobile phase, anodic EOF was obtained using these CSPs (26). In spite of providing rapid separation times, the positively charged polysaccharide CSPs are very time consuming to synthesize often requiring up to at least six or seven different steps. Along the same lines our group have recently synthesized a negatively charged polysaccharide based CSP for CEC and CEC-MS (28).

In this work, the strong cation exchange sulfonated groups (SO₃⁻) that are bonded to the CDMPC were investigated for high throughput analysis. One may question whether a combination of short column packed with CDMPC-SO₃ and outlet side injections would lead to further improvement in sample throughput by CEC. This is because the EOF is generated not only by the silanol groups on the silica gel, but also by the sulfonate groups present on the CDMPC. There-fore, using 7cm effective length with outlet side injections CDMPC-SO₃ was utilized to develop high throughput separations with improved S/N without losing enantioselectivity. To show the advantages of the negatively charged CSP, comparison runs were performed using a neutral underivatized CDMPC column in both CEC-UV and CEC-MS modes.

2. Experimental

2.1. Reagents and materials

The Nucleosil 5μm, 1000Å bare silica gel particles were purchased from Macherey-Nagel (Bethlehem, PA, USA). The HPLC grade solvents acetonitrile (ACN), methanol (MeOH) and ethanol (EtOH), ammonium formate (NH₄COOH), acetic acid (HOAc), formic acid (HCOOH), tetrahydrofuran (THF), isopropanol (IPA) microcrystalline cellulose, 3,5-dimethylphenyl isocyanate, anhydrous pyridine, sulfur trioxide–pyridine complex (SO3·Py), and all of the chiral analytes were purchased from Sigma–Aldrich (Milwaukee, WI, USA). Phosphoric acid (H₃PO₄) and ammonium hydroxide (NH₄OH) were obtained from EMD chemicals (Gibbstown, NJ, USA). Stock solutions of all analytes were prepared as 2 mg/mL solutions in EtOH and working standard solutions were diluted to appropriate final concentrations in the range of 0.5–1.0 mg/mL with triply deionized water. Water used in this study was purified by a Barnstead Nanopure II Water System (Dubuque, IA, USA). Stock background electrolyte solutions (BGE) were prepared by dissolution of the appropriate salt (ammonium formate) in triply deionized water followed by pH adjustment with formic acid. The final mobile phases were then prepared fresh daily by combining volumetric ratios of ACN, water and stock BGE.

2.2. Synthesis of sulfonated cellulose tris (3,5-dimethylphenylcarbamate) (CDMPC-SO₃)

The CDMPC-SO₃ chiral selector was synthesized according to a procedure outlined in our previous work (28). Briefly, the CDMPC-SO₃ was prepared by a sulfonation reaction of CDMPC with SO3·Py (Scheme 1). The synthesis of the CDMPC was based on the reaction described by Okamoto (29). Approximately 3 grams (~20 mmol) of microcrystalline cellulose was weighed out into a round bottom flask along with a clean dry stir bar and allowed to dry under vacuum at 60 °C for 12 h. After drying, 40 mL anhydrous pyridine was added to the round-bottom flask containing the dried cellulose. This mixture was attached to a reflux condenser and allowed to stir and heat at 100 °C for 6 h. After slowly adding 10 grams (61 mmol) of 3,5-dimethylphenyl isocyanate into the flask, the mixture was stirred at 100 °C for 2 days. Next, the mixture was allowed to cool and poured into a large beaker containing 3 L of methanol and allowed to stir for 2 h to precipitate out the product. The product obtained was a white solid that was filtered and dried under vacuum at 60 °C for 12 h. The general yield of the product (CDMPC) was between 93–95%. The CDMPC was then characterized by elemental analysis to confirm the presence of the phenylcarbamate group. Elemental analysis found that the product consisted of C at 64.72%, H at 6.19% and 6.78% N (calculated: 65.44% C, 6.49% H, and 6.94% N). Next, 4 grams (6.6 mmol) of dried CDMPC taken in a round bottom flask was dissolved in 80 mL of anhydrous pyridine. After all of the CDMPC had dissolved, 3.2 grams (20 mmol) of SO3·Py was slowly added dropwise to the round bottom flask and reflux was continued at 100 °C for 2 days. After this, the reaction was allowed to cool and 100 mL of methanol was added to the round bottom flask. Next approximately 70% of the solution (based on observation before and during evaporation) was removed by rotoevaporation. The concentrated mixture was slowly poured from the round bottom flask into a large beaker containing 3 L of methanol and stirred for 1–2 h. The obtained white precipitate was filtered and recrystallized with THF and IPA. This was followed by drying under vacuum at 60 °C for 12 h. The approximate yield of CDMPC-SO3 was 70%. The elemental analysis found that the product consisted of 61.37% C, 5.78% H, 6.78% N and 1.81% S (calculated: 63.43% C, 6.12% H, 6.77% N, 2.0% S).

Scheme 1.

Scheme for the synthesis of CDMPC and CDMPC-SO₃

2.3. Preparation of cellulose phenyl carbamate CSPs

The final CSP was prepared by coating bare silica gel particles with either the CDMPC or CDMPC-SO₃ material according to a procedure described earlier in our previous publication (28). Briefly, 0.1 g of the polysaccharide material was first dissolved in 15 mL of THF in a small round bottom flask. Next, 0.5 g of bare silica gel was added to the flask, sonicated for 10 min to evenly disperse the cellulose based material and the silica gel, followed by roto-evaporation of THF. The CSP was finally dried under vacuum at 60 °C for 12 h before being used to pack columns.

2.4. CEC-UV and CEC-MS column fabrication

The fused silica capillaries (O.D. 375μm × 100μm i.d.), obtained from Polymicro Technologies (Phoenix, AZ, USA) were used to fabricate the columns for CEC-UV and CEC electrospray ionization (ESI)-MS. The packing procedure was the same as the one reported earlier by our group (28), except that packing bed was shortened to 7 cm. After the CSP was packed into the capillary, the CEC-UV column was flushed with water for 2 h and inlet and outlet frits were produced with a home-made frit burner. As stated before, the typical CEC-UV column consists of 7 cm packed bed and 28 cm unpacked segment.

Using the same type of capillary, the CEC-ESI-MS columns were fabricated according to a method previously outlined by our group (30). For the fabrication of all CEC-MS columns, a single inlet frit was used in conjunction with an internal outlet taper. The first step involved the formation of an internal taper at the outlet end of the capillary that is 12μm wide at its narrowest. Next, the internal tapered column was slurry packed at 200 bar with the CDMPC or CDMP-SO₃ CSPs using a Knauer pneumatic pump (Wissenschaftliche Gerätebau, Dr. Ing. Herbert Knauer GmbH, Berlin, Germany). Finally, after the CSP was packed, the column was flushed with water for 2 h and the inlet frit was burned using a home-made frit burner.

2.5. CEC-UV and CEC-MS Instrumentation

The CEC-UV experiments were carried out on an Agilent capillary electrophoresis system (Palo Alto, CA, USA) interfaced to a diode-array detector. The CEC ESI-MS experiments were carried out with a CE instrument interfaced to a single quadrupole mass spectrometer, Agilent 1100 series MSD. An Agilent 1100 series HPLC pump equipped with a 1:100 splitter was used to deliver the sheath liquid. The interface of the CE to MS was made possible by a G1603A CE MS adapter kit and a G1607 CE–ESI-MS sprayer kit also provided by Agilent Technologies. The Agilent ChemStation and CE-MS add-on software (V 10.02) were used for instrument control and calculation of chiral resolution (Rs), retention time (t_r), selectivity (α) and separation efficiency (N). For high sensitivity experiments an Agilent high sensitivity detection cell (Part#: G1600-60027) was used. The installation of the high sensitivity detection cell requires two separate pieces of capillary, an inlet and outlet section of capillary. In our case the outlet section was the 7 cm packed segment of capillary and the inlet section was empty (Figure 1). Before joining the segments to the high sensitivity flow cell, the packed outlet section of capillary was flushed with mobile phase using the HPLC pump and the empty inlet section of the capillary was flushed with mobile phase using a syringe. The two sections were then carefully joined to the high sensitivity flow cell using the finger tight plastic screws provided with the flow cell. Finally, the attached pieces were placed in a traditional CEC-UV cartridge.

Figure 1.

Cartoon illustrating the method of short column outlet side injections. The arrow represents the direction of the EOF.

2.6. CEC-UV and CEC–ESI-MS conditions

All of the mobile phases and sheath liquids were sonicated for 15 min, filtered using a vacuum filter flask (47 mm, 0.45 μm filters) and degassed for 10 min before use. The columns were preconditioned using the Knauer pneumatic pump with the desired mobile phase at 180 bar for 2 h. This was followed by further voltage conditioning on the CEC instrument for another 2 h to reach the desired running voltage. Injections were performed through high flushing mode of the Agilent CE (applying 6 bars and −6 kV for 6 sec). To perform the outlet side injections as illustrated in Figure 1, a negative voltage was applied (specified in the Agilent ChemStation software Injection submenu) while the injection vial was directed by the software to the outlet side of the capillary. Moreover, during the CEC-UV separation, voltage was set at −20 kV to drive the mobile phase from the outlet buffer vial toward the inlet vial. This is the opposite polarity of voltage that is normally used when traditional CEC-UV is carried out. During the separation, an external pressure of 6 bar was also applied to the inlet and outlet buffer vials to suppress bubble formation. For CEC–ESI-MS, the separation voltage was set at +20 kV along an external pressure of 12 bar applied to the inlet vial. Because outlet side injections cannot be carried out with CEC-ESI-MS, no reversal of the voltage was necessary was needed in CEC-UV experiments. The following spray chamber conditions were used unless otherwise stated: sheath liquid, MeOH/H2O (90:10, v/v) containing 50 mM NH₄OAc, flow rate, 5.0 μL/min; capillary voltage, +3000 V; fragmentor voltage, 80 V; drying gas flow rate, 5 L/min; drying gas temperature, 130 C; nebulizer pressure, 4 psi. The selective ion-monitoring (SIM) mode was set at corresponding polarity to monitor the desired protonated ([M+H]⁺) or deprotonated ([M H])molecular ions for each investigated chiral analyte.

3. RESULTS AND DISCUSSION

3.1. Effect of packed-bed length in CEC-UV

Because CEC generates efficiencies approximately three to four fold greater than HPLC (31), it should be possible to perform rapid chiral separations on short CEC column and still generate the acceptable resolution. This is especially true for CSP, which has higher chiral selectivity so that some drop in efficiency (due to drop in packed bed length) still makes it possible to generate baseline resolution of closely eluting enantiomers. To determine the feasibility of reduction in analysis time, short packed bed with a 7 cm column and outlet side injection was compared with a conventional 20 cm column and inlet side injection. A comparison of the chromatograms using both CDMPC (Figure 2A and C) versus. CDMPC-SO₃ (Figure 2 B and D) CSPs for the same analyte under identical mobile phase conditions was carried out. For a mixture of (±)-warfarin (WAR), back-to-back runs were conducted on both 20 cm column (top chromatograms, Figure 2 A–B) with inlet side anodic injection and on a 7 cm column (bottom chromatograms, Figure 2 C–D) using outlet side cathodic injection. With the CDMPC CSP, both the 20 cm and 7 cm column were able to reach better than baseline resolutions (3.2 for 20 cm columns vs. 3.0 for the 7 cm column). The advantage lies in the reduced separation time with the short 7 cm column because it was able to provide excellent resolution (Rs = 3.0) in less than 4 min. In comparison, the long 20 cm column provided nearly equivalent resolution (Rs = 3.2), but separation time was over 11 min. Note, there was very little difference between the selectivities for these two columns, though there was some drop in the efficiency for the 7 cm column.

Figure 2.

Comparison of a 20 cm cellulose based CSP column with standard inlet side injections to a 7 cm column packed with outlet side injections for CEC-UV of anionic racemates, (±)-warfarin (WAR). 35 cm × 100 μm i.d. capillary; packed with 5 μm 1000 Å bare silica coated with 20% CDMPC or CDMPC-SO₃. (A) conventional chiral CEC performed with CDMPC as CSP, mobile phase of 70% ACN 5 mM NH₄COOH, pH 3.5, +20 kV, 25°C, 214 nm, with an electrokinetic injection +6 kV, 6 s; (B) Conventional chiral CEC performed with CDMPC-SO₃ as CSP, mobile phase, applied voltage and injection conditions are the same mention as (A); (C-D) “Short-end injection” chiral CEC performed with CDMPC and CDMPC-SO₃ CSP, respectively with applied voltage −20 kV and electrokinetic injections of −6kV, 6 s For all other conditions see the experimental section. Efficiencies expressed as plates/m.

As expected, both long and short packed bed containing CDMPC-SO₃ as CSP had even faster separations with higher efficiency when compared to their CDMPC counterpart of the same packed bed length. This is due to the presence of the charged groups on the CDMPC-SO₃ providing an enhancement of the EOF. For example, the CDMPC-SO₃ column that was packed with only 7 cm of CSP (Figure 2 D) showed the fastest separation in less than 2 min for (±)-WAR. With CDMPC-SO₃ as packed bed, both lengths of column were still able to provide better than baseline resolution for (±)-WAR. However, the 7 cm column had more than three-fold decrease in separation time with a slight drop in resolution (Rs = 1.9, analysis time ~2 min Figure 2D) when compared to the 20 cm column (Rs = 2.4, analysis time ~7.5 min, Figure 2B). Again, there was very little difference in the selectivity with a drop in the efficiency for the 7 cm column as compared to the 20 cm column. This is acceptable based on the baseline separation, higher S/N and very rapid analysis time. These experiments demonstrated the combination of CDMPC-SO₃ as stationary phases and short-end injection provides the best CEC-UV conditions reported to-date for the rapid and sensitive analysis of enantiomers of (±)-WAR.

3.2. CEC-UV screening of chiral compounds with CDMPC and CDMPC-SO₃

Once the applicability of the short column with outlet side injection design had been established, multiple chiral compounds were screened using the same set up beginning with mobile phase conditions optimized in our previous study (28). The first set of analytes that were looked at included racemic mixtures of aminoglutethimide (AG) and 2,2,2-trifluoro-1-(9-anthryl) ethanol (TFAE). When the positively charged (±) AG (see inset of top chromatogram for structure, Figure 3) was run using the CDMPC based CSP, the analyte was resolved in less than 3 min with better than baseline resolution (i.e., Rs = 2.3). Under the same conditions using the CDMPC-SO₃ stationary phase, the (±) AG was still better than baseline resolved (i.e., Rs = 1.9) in less than 2 min. The drop in resolution and separation time of (±) AG can best be explained by the CDMPC-SO₃ CSP having negative charges available for generation of faster EOF as compared to the CDMPC CSP. In turn, the greater EOF leads to quicker mobility of the ions and a faster analysis. As shown in Figure 3, there was very little drop in α, but the S/N was higher and very little change in N of (±) AG was observed using CDMPC-SO₃ compared to CDMPC CSPs.

Figure 3.

Chromatographic examples of the CEC-UV chiral separations of (±)-AG (A and B) and (±)-TFAE (C and D) using short column with outlet side injections packed with CDMPC or CDMPC-SO₃. 35 cm × 100 μm i.d. capillary; 7 cm packed with 5 μm 1000 Å bare silica coated with 20% CDMPC or CDMPC-SO₃ M.P. 70% ACN 5 mM NH₄COOH, pH 3.5, −20 kV, 25°C, 214 nm. Electrokinetic injections performed at −6 kV, 6 s. For all other conditions see the experimental section. Efficiencies expressed as plates/m.

The inset in bottom chromatograms (Figure 3) shows the structure of (±)TFAE. Comparison of CDMPC and CDMPC-SO₃ for the separation of the neutral racemate (±) TFAE at pH 3.5 is shown in Figure 3 C-D. As was the case for (±) AG, very similar trend is seen for Rs and separation time for the (±)TFAE chiral analyte. The CDMPC was able to provide better resolution of the analyte but relatively longer run time (Rs = 3.1, run times ~ 4.5 min) compared to CDMPC-SO₃. However, CDMPC-SO₃ based column showed a better trade-off between separation time and enantioresolution (Rs = 2.1, run time ~ 2.5 min). Note, that unlike the enantiomers of (±) AG, the N of (±) TFAE was more than doubled but there was a slight increase in S/N and a slight drop in α with CDMP-SO₃ compared to CDMPC CSP.

The next set of chiral analytes screened included racemic glutethimide (GL) and benzoin (BZN) (see inset of Figure 4 for structures). With these two analytes it was necessary to make slight adjustments to the organic solvent composition to see baseline resolution with both the CDMPC and the CDMPC-SO₃ CSPs. The CDMPC as expected provided longer retention times as compared to the CDMPC-SO₃. Therefore, it was able to baseline resolve both GL (Rs = 1.5) and BZN (Rs = 1.7) using the originally optimized 70% ACN containing mobile phase reported in our previous work [28]. On the other hand, CDMPC-SO₃ was able to provide close to baseline resolution for GL (Rs = 1.4) but less than baseline resolution for BZN (Rs = 0.94) using the same concentration of organic solvent. For this reason, the ACN concentration was lowered to 50% to increase the hydrophobicity of the CSP causing the analytes to retain longer on both phases. This longer interaction time provided baseline resolution for both GL (Rs = 1.7) and BZN (Rs = 2.2). Nevertheless, comparing the chromatograms shown in Figure 4 A vs. 4 D and Figure 4 E vs. 4 H it is evident that the CDMPC-SO₃ CSP showed the best overall enantioseparations (both Rs and a) for both GL and BZN with very little adjustment in mobile phase conditions. Therefore, CDMPC-SO₃ was considered the better of the two stationary phases examined.

Figure 4.

Chromatographic examples of CEC-UV chiral separations of (A-D) (±)-glutethimide (GL) and (E-H) (±)-benzoin (BZN) using short column outlet side injections packed with CDMPC or CDMPC-SO₃. 35 cm × 100 μm i.d. capillary; 7 cm packed with 5 μm 1000 Å bare silica coated with 20% CDMPC or CDMPC-SO₃; M.P. 50% or 70% ACN 5 mM NH₄COOH, pH 3.5, −20 kV, 25°C, 214 nm. Electrokinetic injections performed at −6 kV, 6 s. For all other conditions see the experimental section. Efficiencies expressed as plates/m.

The retention order of the chiral compounds on the two different CSPs does not follow the lution order expected solely based on charge or hydrophobicity alone. Based on pKa and logP data, the first eluted compound would be expected to be positively charged AG, followed by the neutral GL, BZN, TFAE and lastly the negatively charged WAR. The actual order of elution from first eluted to last eluted chiral compound on CDMPC CSP is: AG<GL<BZN<WAR< TFAE, which seem to be solely based on hydrophobicity. The order of elution on the CDMPC-SO₃ is: AG<BZN<WAR<TFAE< GL. Evidently a complex combination of electrostatic and hydrophobic effects is controlling the retention order of the same four chiral analytes using CDMPC-SO₃ CSP in CEC.

The column-to-column repeatability was tested with the separation of the ±-AG enantiomers examining retention time, resolution, and efficiency across 50 runs on three different columns. The percent relative standard deviations (%RSD) for retention times of the two enantiomers of ±-AG for all three columns were between 2.5% and 4.6%, with intercolumn %RSD of 5.8% and 6.4% for peak 1 and peak 2, respectively. For all three columns, the %RSD for Rs was between 3.5% and 4.6%, with intercolumn %RSD of 6.9. Finally, the average efficiency for all three columns had %RSD between 5.8% and 6.2%, with intercolumn values of 7.4%. The data tabulated in Table 1 demonstrates that the short column outlet side injection strategy is stable and robust with respect to repeatability of retention time, resolution, and peak efficiency.

Table 1.

Intra- and inter-column repeatability of the short colum outlet injection set-up for average retention time (tr_avg), average resolution (Rs_avg) and efficiency (N_avg) of (±AG) with CDMPC-SO₃⁻ stationary phase

	travg, %RSD		Rsavg, %RSD	Navg, %RSD
	peak 1	peak 2
Column 1 n = 50	2.5	3.6	4.6	6.2
Column 2 n = 50	3.1	4.6	3.5	5.8
Column 3 n = 50	4.2	4.3	4.3	6.1
n = 150	5.8	6.4	6.9	7.4

3.3. CEC-UV with high sensitivity flow cell detection

To increase sensitivity of chiral CEC-UV, tests were run with a high sensitivity flow cell using (±)-WAR as a chiral test compound. As shown in Figure 5B, the use of the high sensitivity detection cell provided nearly ten fold increase in S/N (S/N = 500) when compared to the same run using a standard detection cell (Figure 5A). Unfortunately, due to the design of the high sensitivity flow cell some band broadening occurs, accounting for some drop in resolution and efficiency. However, such band broadening could be easily offset by adjustment of the mobile phase conditions and data acquisition parameters.

Figure 5.

Comparison of short column outlet side injections using a standard cell (A) and a high sensitivity flow cell (B) for detecting warfarin using a CDMPC-SO₃ CSP. Column dimensions: 35 cm × 100 μm i.d. capillary; 7 cm packed with 5 μm 1000 Å bare silica coated with 20% CDMPC-SO₃ M.P. 70% ACN 5 mM NH₄COOH, pH 3.5, −20 kV, 25°C, 214 nm. Electrokinetic injections at −6 kV, 6 s. For all other conditions see the experimental section. Efficiencies expressed as plates/m.

The high sensitivity detection cell with the short column outlet side injection strategy was applied to the analysis of (R/S)-WAR enantiomeric ratio. Using the high sensitivity detection cell it was possible to distinguish the differences in the enantiomeric ratios of R- and S-WAR at various concentrations (data not shown). The calibration curves were established in the range of 0.5–1000 μg/mL for the R- and S-enantiomer in blank plasma sample spiked with (0±)-WAR. The % recovery in blank plasma samples spiked at various concentrations for R- and S-enantiomer of WAR were calculated by comparing the peak areas of blank plasma sample to standard solutions and was in the range between 89%-94% (data not shown).

3.4. Effect of packed-bed length in CEC-ESI-MS

As with the studies done with CEC-UV, the first step was to evaluate if the effect of reducing the effective column packed bed length from 20 cm to 7 cm will have the same effect in CEC-MS. Here (±)-WAR is studied as a model test racemic analyte using only the CDMPC-SO₃ CSP in determining the feasibility of performing CEC-ESI-MS studies. As shown in Figure 6, conventional 20 cm packed bed using 70% ACN was able to close to baseline resolve (Rs = 1.4) the enantiomers of (±)-WAR in 23 min. However, when the short 7 cm packed bed column was used, no baseline resolution of (±)-WAR was seen with a mobile phase containing 70% ACN (data not shown). Once the ACN was decreased to 50%, close to baseline resolution (Rs = 1.3) was achieved under 14 min. As expected, an increase in band broadening was seen with the CEC-ESI-MS runs as shown by the decreases in efficiency (Figure 6) when the data from the CEC-UV mode using CDMPC-SO₃ (Figure 2) is compared. For the 20 cm column, there was a slight drop in N from 22,000 plates/m observed in CEC-UV to 20,000 plates/m in CEC-MS. Similar trend was seen for the 7 cm column, which showed N to drop from 9,000 plates/m down to 8,000 plates/m. This is one of the major limitation with CEC-MS, which requires a total length of at least 50 cm for the CE capillary to reach the MS detector. Therefore, this additional open tube length seems to contribute to some band broadening.

Figure 6.

Comparison of (A) 20 cm column with (B) 7 cm column packed with CDMPC-SO₃ CSP in CEC-ESI-MS. Column dimensions: 50 cm × 100 μm i.d. capillary; 7 cm or 20 cm packed with 5 μm 1000 Å bare silica coated with 20% CDMPC-SO₃. The mobile phase contains various %ACN, 5 mM NH₄COOH, pH 3.5, 20 kV, 25°C, electrokinetic injection at 10 kV, 10 s. SIM of WAR in negative ion mode [M-H]⁻ = 307 m/z; drying gas flow rate: 5 L/min, nebulizer pressure: 4 psi; drying gas temperature: 130°C. Efficiencies expressed as plates/m.

3.5. CEC-ESI-MS of Chiral Compounds with CDMPC and CDMPC-SO₃

As mentioned and shown earlier, because of the instrumental constraints of using a longer (i.e., at least 50 cm) column in CEC-ESI-MS mode, it was predicted that there would be an increase in band broadening with packed and open segments in the column, leading to poorer efficiency with this detection mode. Furthermore, it was expected that the CDMPC-SO₃ stationary phase would provide better separations in CEC-ESI-MS as well, although, both types of CSP were examined in the CEC-ESI-MS study for comparison. Even though there was some loss in efficiency and resolution, the losses were not considered a significant detriment to preclude testing the two polysaccharide CSPs in CEC-ESI-MS mode. Hence, the same chiral analytes were profiled as those screened in the CEC-UV mode except for BENZ, which showed little to no chiral resolution under any of the mobile conditions tried in the CEC-ESI-MS study.

Because (±)-WAR was not completely baseline resolved in the column length study, (±)-WAR and (±)-GL were screened at various amounts of ACN as displayed in Figure 7. The CDMPC and CDMPC-SO₃ CSPs were compared in back-to-back runs. As expected and shown with the screenings of chiral test analytes using CEC-UV, there was an increase in retention time and band broadening with the decrease in % ACN. With the CDMPC column baseline resolution (Rs = 1.5) of (±)-WAR was reached in approximately 18 min using 50% ACN (Figure 7A). However, the CDMPC-SO₃ column did show close to baseline resolution (Rs = 1.3) for the same analyte in just under 14 min (Figure 7B). Because the CDMPC-SO₃ column provided less than baseline resolution for (±)-WAR, the percentage of ACN was decreased further from 50% to 40%. Decreasing % ACN on the CDMPC column showed increased resolution (Rs = 3.4) but also increase separation time (~24 min) as one would expected from the increase in polarity of the mobile phase (Figure 7C). The CDMPC-SO₃ stationary phase was finally able to reach a high resolution (Rs = 2.3) with the same mobile phase composition of 40% ACN but also with shorter run time of 18 min (Figure 7D). For the (±)-GL, only a mobile phase containing 30% ACN was able to show any baseline resolution for either column. The CDMPC CSP provided a resolution of 1.5 in just less than 35 min (Figure 7E), while the CDMPC-SO₃ column showed higher resolution of 1.7 but shorter separation time of less than 27 min and close to an order of magnitude higher S/N (Figure 7F). As expected, efficiencies of all chiral analytes usually drop due to the increased band broadening seen in CEC-ESI-MS, although they still remains at acceptable levels to allow baseline resolution.

Figure 7.

Chromatographic examples of CEC-ESI-MS chiral separations of (±)-WAR and (±)-GL using a short 7 cm column packed with CDMPC or CDMPC-SO₃. 50 cm × 100 μm i.d. capillary; 7 cm packed with 5 μm 1000 Å bare silica coated with 20% CDMPC or CDMPC-SO₃ M.P. various ACN 5 mM NH₄COOH, pH 3.5, 20 kV, 25°C, electrokinetic injection at 10 kV, 10 s. SIM of molecular ion of deprotonated WAR [M-H]⁻ = 307 m/z and protonated GL [M+H]⁺ = 218 m/z; drying gas flow rate: 5 L/min, nebulizer pressure: 4 psi; drying gas temperature: 130°C; Efficiencies expressed as plates/m.

The (±)-AG and (±)-TFAE were next examined using the short CEC-ESI-MS columns. The same trends seen with (±)-WAR and (±)-GL were also seen in Figure 8 with (±)-AG and (±)-TFAE. Increasing the polarity of the mobile phase was again required to baseline resolve the compounds on both CSPs. When (±)-AG was separated on the CDMPC column at 40% ACN, it was better than baseline resolved (Rs = 2.4) under 19 min (Figure 8A). This resolution was slightly deteriorated (Rs = 2.3) by decreasing the ACN further with the expected increase in retention time (~38 min), which causes peak broadening (Figure 8 C). When the CDMPC-SO₃ column was used, low resolution (Rs = 0.78) was achieved in just less than 16 min (Figure 8B). The drop in the ACN to 30 % led to better than baseline resolution (Rs = 2.2) in less than 28 min for the CDMPC-SO₃ CSP (Figure 8 D). The (±)-TFAE experienced the same trend as the (±)AG (data not shown) except that in this case very little mobile phase adjustment was needed. For both CDMPC and CDMPC-SO₃, baseline resolution was achieved using 50% ACN in the mobile phase (CDMPC: Rs = 1.6 in ~21min; CDMPC-SO₃: Rs = 1.9 in ~17 min) (Figure 8 E-F).

Figure 8.

Chromatographic examples of CEC-ESI-MS chiral separations of (±)-aminoglutethimide (AG) and (±)-TFAE using a short 7 cm column packed with CDMPC or CDMPC-SO₃. 50 cm × 100 μm i.d. capillary; 7 cm packed with 5 μm 1000 Å bare silica coated with 20% CDMPC or CDMPC-SO₃ M.P. various ACN 5 mM NH₄COOH, pH 3.5, 20 kV, 25°C, electrokinetic injection at 10 kV, 10 s. SIM of protonated molecular ion, AG [M+H]⁺ = 233 m/z and deprotonated molecular ion, TFAE [M-H]⁻ = 275 m/z; drying gas flow rate: 5 L/min, nebulizer pressure: 4 psi; drying gas temperature: 130°C; Efficiencies expressed as plates/m.

4. Conclusions

In this study, we have illustrated the benefits of combining short-packed bed column and outlet side injection containing either CDMPC or CDMPC-SO₃ CSP in CEC-UV. In the case of all chiral analytes (except benzoin), the CEC-UV with the CDMPC-SO₃ CSP provided higher throughput with baseline resolution or better when compared to the CDMPC CSP. The convenience of improving limit of detection is demonstrated using a short-packed bed outlet injection with high sensitivity UV-flow cell for the analysis of R/S ratios of an important chiral drug, warfarin. Therefore, combining outlet side injection method, a short column packed with CDMPC-SO₃ CSP and a high sensitivity flow cell has a great potential for the analysis of other pharmaceutical drug in biological samples in CEC-UV.

While in CEC-MS outlet-end injection is not feasible, the use of short 7-cm packed bed provided significantly lower separation time compared to the standard 20-cm column length reported earlier in the literature for CEC-MS. Furthermore, the combined use of a short 7-cm packed bed of CDMPC-SO₃ still provided higher throughput compared to CDMPC CSP, except in the case of benzoin. Nevertheless, this high throughput approach provided several folds faster separation time for chiral compounds with baseline resolution. The differences between CEC-UV and CEC-ESI-MS modes is that these two modes provide complementary information. The CEC-UV mode has shorter separation times and higher resolution, while CEC-MS mode provided greater sensitivity and the possibility for mass and structural information. In addition, the concept of using these short chiral CEC columns is ideal for quick screening of new commercial chiral packings or high throughput screening of combinatorial library of chiral compounds to find the lead as well as to determine structure-enantioselective relationships.