Use of Achiral Columns Coupled with Chiral Columns in SFC Separations to Simplify Isolation of Chemically Pure Enantiomer Products

• Amgen Inc.

Introduction

Often pharmaceutical intermediates require chiral purification by SFC after a reaction step to isolate enantiomers of divergent biochemical activity[1-4]. Following a reaction, achiral impurities are always present at some level relative to the desired product, often significant even after normal phase “clean up” separation in advance of chiral purification. Frequently, conditions with one chiral stationary phase (CSP) among an SFC screening set are found that separate the enantiomers and concurrently separate interfering achiral impurities such that an efficient scale up method is possible with one chiral column. In many cases however, co-eluting or closely eluting impurities can prevent efficient purification from being achieved in a single step. In cases of insufficient resolution from achiral impurities with one CSP, one or both enantiomers collected then require a second stage of purification. A separate mobile and stationary phase process is necessary to achieve an end product of adequate chiral purity and chemical purity.

If a stationary phase exists that yields the needed achiral selectivity with the same mobile phase used for a chiral separation method, it follows that the coupling of these phases can yield selectivity for all components in one chromatographic method [5]. Achiral and chiral columns have been coupled for preparative separations of mixtures containing interfering impurities with the racemate of interest. Alexander et al. described a separation of the stereoisomers of cinnamonitrile and hydrocinnamonitrile intermediates using a silica column coupled with a Cellulose tris (3,5-dimethylphenyl carbamate) chiral column [6]. Zeng et al. designed a sophisticated 2D SFC/SFC/ MS purification system with modifications to a mass-directed prep LC and custom software for control [7]. Using mass-directed fractionation [8, 9], the racemate peak is isolated in the first dimension using a 2-ethylpyridine achiral column, and then eluted from a trapping column with the mobile phase for the second dimension onto a chiral column for enantiomer resolution.

If the necessary achiral selectivity does not occur with the single SFC achiral phase, the benefits of these processes are lost. Since no universal SFC achiral column exists similar to C18 for reversed-phase [10], a screening process of candidate columns can be employed. An apparatus for automated coupling of analytical columns SFC system by Welch et al. [11] has been described for coupling of chiral column pairs. A similar setup for coupled chiral-achiral screening can likewise be employed. A suitable pre-purifi cation analysis resulting from this arrangement can lead to a single step isolation of enantiomers from achiral impurities in a coupled column preparative step.

Results and Discussion

Achiral Column Set Selection Experiment

A general gradient analysis performed with an analytical SFC/ MS system on eight achiral stationary phases was used to plot the retention data of a set of 70 diverse pharmaceutical compounds on a standard test plate. The phases tested were Silica, HILIC (cross-linked Diol), Nitro, DEAP (diethylaminopropyl), Pyridyl Amide, Imidazole, 4-Ethylpyridine and 2-Ethylpyridine, all 4.6 x 100 mm, 5 micron. The linear gradient program used for this achiral column evaluation was: 5 – 60% modifi er (methanol with 20mM NH3) in 3 min., held at 60% for 0.67 min., lowered to 5% over 0.33 min., and held for re-equilibration at 5% for 1 min. The columns were installed in various combinations on a software controlled 6-channel column selector on board an external autosampler for this experiment (see Figure 1). The SFC system outlet pressure was fi xed at 100 bar, column temperature 40⁰C, and total fl ow rate was 4.0 mL/min.

Figure 1. Coupled column system diagram. Flow through each of two column channels in series is possible with this fl exible setup on the SFC/MS analytical system. Bypass lines are included in each column switching valve to allow for single column analyses as needed.

Assembling all retention times from the 70 compounds yields a frequency distribution used for comparison of the various achiral phases tested. Shown in Figure 2 is the distribution found for six of the columns studied. (Two of the other phases tested had many cases of standards not eluting, 15 in the case of Nitro, four for DEAP. These were deemed unsuitable for the application and are not represented in this fi gure.) Ideal properties for columns to be included in a general screening set are differences among compounds’ retention from other columns in the set, relatively few cases of over-retention, t_R > 4 min. with this gradient, and a limited number under-retained, t_R < 2 min. These properties increase the likelihood of fi nding alternative selectivity among the columns for arbitrary compounds to be encountered in the future.

Figure 2. Histograms of retention times obtained from SFC gradient analysis of 70 diverse standard pharmaceutical compounds on each of six achiral columns arranged in 0.5-minute wide bins.

In Figure 2, the histogram for the 2-ethylpyridine column “2-EP” exhibits overall shorter retention for the set shifting the retention distribution signifi cantly to the left compared with those of other phases. Although 2-ethylpyridine is one of the most widely used for achiral SFC, this data deems it a poor choice to include in an achiral screening set since it implies that two random compounds are more likely to have similar retention. The column is thus less likely to produce the added selectivity one may need in a complementary separation for this application. Since Pyridyl Amide and 4-ethylpyridine “4-EP” exhibited the most similarity among specifi c compound retentions in the study, Pyridyl Amide was chosen to be kept in a fi nal four-column screening set (Silica, HILIC, Imidazole and Pyridyl Amide) as 4-EP added the least diversity to the group.

Coupling Different achiral phases with one Csp

Having determined this set of achiral columns, each can be screened by coupling with the chiral column and mobile phase suitable for the racemate separation. For example, if a good chiral separation was achieved with 20% methanol modifi er on CSP #1, then achiral columns 1, 2, 3 and 4 would be individually coupled with that CSP and run sequentially with the 20% isocratic method, simply doubling the analysis time. The data from each of the four coupled methods would be compared to fi nd the most favorable separation conditions accounting for both resolution of the impurity from enantiomer peaks as well as peak shapes resulting from the two-column separation.

To illustrate the case of a chiral separation with an interfering achiral impurity, a racemate, Chlormezanone, was spiked with Chrysin and analyzed with coupled chiral and achiral columns. The recommended process for a screen to achieve a one-step chiral/achiral separation is performed by using conditions from the optimal isocratic chiral method (with acquisition time doubled from 5 to 10 minutes) applied to each of the four different achiral columns in series with an Amylose chiral column (see Figure 3). The retention of enantiomers and that of the achiral compound are aff ected differently by the achiral column upstream of the CSP. While chiral selectivity is maintained after elution from the coupled columns, the unique selectivity of each achiral column results in a different chromatographic profi le. Figure 3 data for the Nitro column shows Chlormezanone and Chrysin both retained more strongly than with other achiral columns applied, Chrysin so strongly retained it does not elute with the same isocratic method. DEAP also strongly retains Chrysin although it elutes within the 10-minute run time. With the 4-EP and Imidazole columns, Chrysin is less retained but coelutes with the second-eluting enantiomer of Chlormezanone. However with the HILIC column coupled in this example, there is good selectivity for all three peaks. With a suitable preparative scale up, enantiomers of Chlormezanone may be effi ciently isolated free from contamination with Chrysin.

Figure 3. Comparison of 5 different achiral columns coupled with Amylose tris [(S)-α-methylbenzylcarbamate for chiral/ achiral separation of Chlormezanone with Chrysin “impurity” using isocratic method 25% ethanol (20 mM NH3) / 75% CO₂ at 4 mL/min.

Coupled Column screen with synthetic racemate and achiral impurity

The result of a case in which the achiral/chiral column screen was performed is illustrated in Figure 4. A signifi cant achiral impurity was present in this sample and insuffi cient separation was found in chiral screening. From the screen, the best isocratic method for chiral resolution was 20% methanol (20 mM NH₃ additive) + 80% CO₂ on a 4.6 x 100 mm immobilized Amylose 3,5-dimethylphenyl carbamate column. Using these conditions and coupling each of the four selected achiral phases with the CSP, an advantageous selectivity of enantiomers from the achiral impurity was sought. The zoomed-in chromatogram view Figure 4 reveals varying selectivities for Silica, Pyridyl Amide and HILIC, but none completely resolving the impurity from enantiomers. However the bottom chromatogram representing a coupling of the imidazole phase with immobilized Amylose reveals baseline resolution of both enantiomers from the impurity, the latest eluting peak. This is exactly the result sought from the extra screening step allowing the purifi cation of these enantiomers to be achieved in one preparative step.

Figure 4. Comparison of 4 different achiral columns coupled with immobilized Amylose 3,5-dimethylphenyl carbamate for chemistry chiral sample with large impurity using the isocratic method 20% methanol (20 mM NH3) / 80% CO₂ at 4 mL/min.

preparative SFC Comparison – Coupled vs. two-stage separation

To illustrate the impact of coupling columns at the preparative stage, the racemic pharmaceutical compound Diperodon was spiked or “contaminated” with 10% caff eine by weight. The coupled column analytical screening indicated an achiral resolution was possible at conditions which resolved the enantiomers, so a preparative scale SFC experiment with relevant columns was carried out. At the top of Figure 5 a preparative mode stacked injection sequence [12] designed to achieve a fast cycle chiral resolution for the mixture is shown. It is accepted that no achiral resolution would be achieved in this step so with favorable chiral separation conditions for an immobilized Cellulose tris (3,5)-dimethylphenyl carbamate column, the caff eine peak mostly overlaps with the front edge of the fi rst eluting enantiomer using this method (see Figure 5). The collected fraction 1 required evaporation after collection and another preparative process to resolve caff eine from enantiomer 1 isolated in the chiral separation step. On the right in the fi gure is the achiral repurifi cation trace of fraction 1 which was required following the chiral-only separation initially performed.

Figure 5. Preparative SFC stacked injection separation traces of Diperodon plus caff eine with chiral column only (upper left), coupled achiral + chiral column (lower left), and fraction 1 repurifi cation chromatogram (right), necessary after insuffi cient separation from caff eine in the fi rst, chiral-only stage.

All the time taken for the second step is unnecessary utilizing the coupled column concept in this example. Using these same mobile phase conditions and cycle time, but with a 15 cm Pyridyl Amide column coupled with a 20 mm length of 0.020” stainless steel tubing to the 25 cm immobilized Cellulose column, the chromatographic data shown lower in the fi gure was obtained. Here the caff eine peak is fully resolved from enantiomer peaks within the three-minute cycle time. With this method, caff eine as well as each enantiomer peak were resolved in one preparative step. Four injections completed the equivalent small quantity sample purifi cation in both cases. The coupled mode process was completed in 53 fewer minutes than the standard two-stage process in this example.

Notwithstanding this case, for any sample showing chiral and achiral selectivity with a coupled column method, relevant data and sample properties must be assessed to determine the likely feasibility of coupled-mode preparative separation. If the cycle time relative to the chiral-only separation is signifi cantly longer, considering this difference and sample size will determine whether the coupled process or twostage process is more time-consuming. Stacked injection processing (purifi cation) is usually the most time-consuming component unless a sample is small. The coupled-mode process also includes time to setup and run the achiral column screening component following chiral isocratic method development. Additional time-consumers in the two-stage process may include additional analytical method development time if, for example, reversed-phase is required to remove the interfering impurity. Following both stages there is additional time also needed to evaporate the fractions obtained.

Conclusions

The methodology described provides a procedure for advantageously isolating chemically pure enantiomers in a readily accessible way for most chiral separations laboratories using SFC. This experiment described a selection process to fi nd a useful, complementary set of achiral columns to facilitate one-step chiral and achiral purifi cation for chemically impure racemates. Other columns than those described here may be useful depending on the compounds being separated so any lab attempting this may need to make a relevant selection. We have found for our laboratory that in roughly 20% of impure chiral samples we can fi nd a coupled separation method which eliminates the need for a two-stage purifi cation. Barriers to more general application include many cases of multiple achiral impurities in some samples as well as the limitation in achiral selectivity under mobile phase conditions necessary to achieve the chiral separation. In this light, it is interesting to consider the possibilities available if additional fl exibility and technology advancements could be incorporated. More diverse achiral selectivity may be accessible with new achiral phases available for SFC. Further work in automated optimization of separations through varying achiral and chiral column lengths with appropriate method development and modeling should also benefi t this application.

Acknowledgements

The author acknowledges the eff orts contributed by Dhanashri Bagal and Brent Murphy enabling these experiments to be performed and reported. The author also acknowledges Wolfgang Goetzinger for test compound provision, as well as Larry Miller and Kyung Gahm for editorial support.

Author Biography

Manuel Ventura, Ph.D., is a Principal Scientist and leader of the Separations group supporting Therapeutic Discovery Chemistry at the South San Francisco site of Amgen, Inc. He began his career at Pfi zer, Inc. where he was key developer of a novel SFC/MS interface applied to high-throughput library analysis. Dr. Ventura has continued developing new platforms and processes utilizing SFC throughout his career and has contributed numerous presentations and publications. Manuel received his Ph.D. in analytical chemistry from the University of Texas at Austin in 1997.

References

1. White., C.J. J. Chromatogr. A, 1074, 163 (2005).
2. Welch, C.J. (2007) Chiral Chromatography in Support of Pharmaceutical Process Research, in Preparative Enantioselective Chromatography, Blackwell Publishing Ltd, Oxford, UK (Ed. G. B. Cox.)
3. Miller, L., Potter, M. J. Chromatogr. B, 875, 230 (2008).
4. De Klerck, K., Mangelings, D., Vander Heyden, Y. J. Pharm. Biomed. Anal. 69, 77-92 (2012).
5. Phinney, K.W., Sander, L.C., Wise, S.A. Anal. Chem., 70, 2331-2335 (1998).
6. Alexander, A.J., Staab, A. Anal. Chem., 78 (11), 3835-3838 (2006).
7. Zeng, L., Xu, R., Zhang, Y., Kassel, D.B. J. Chromatogr. A, 3080-3088 (2011).
8. Wang, Y., Xue, X., Xiao Y, Zhang, F., Xu, Q. Liang, X. Rapid Commun. Mass Spectrom., 15(22), 2067-75 (2001).
9. Zhang, X., Towle, M.H., Felice, C.E., Flament, J.H. and Goetzinger, W.K. J. Comb. Chem., 8, 705-714 (2006).
10. McClain, R., Przybyciel, M. LCGC Asia Pacifi c, 15(4), 26-33 (2012).
11. Welch, C.J., Biba, M., Gouker, J.R., Kath, G., Augustine, P., Hosek, P. Chirality, 19, 184- 189 (2007).
12. Berger, T.A., Smith, J., Fogelman K., Kruluts, K. Am. Lab., 34, 14-20 (2002).