Todd:PZQ Student Optimization

Home        Contact        Internal        People        Papers        Research        Teaching        Links        News        Construction       

Student-led Optimization of the Resolution of Praziquantel

New authors - add yourselves here in alphabetical order:

Charles T. Cox, Department of Chemistry, Stanford University, Stanford, CA 94305, USA
Bojan Milic, Departments of Chemistry and Biology, Stanford University, Stanford, CA 94305, USA
Piero Olliaro, Special Programme for Research and Training in Tropical Diseases (TDR), World Health Organization, Avenue Appia 20, 1211 Geneva 27, Switzerland
Matthew H. Todd, School of Chemistry, The University of Sydney, NSW 2006, Australia
Michael Woelfle, School of Chemistry, The University of Sydney, NSW 2006, Australia

Authors have contributed to the open electronic lab notebook where the original research was first disclosed.

Abstract

A preparation of enantiopure praziquantel was recently discovered that involves the classical resolution of a hydrolysis product, praziquanamine. The protocol is experimentally simple, and involves no chromatography. For the procedure to be of the greatest impact, the price must be kept very low, meaning that the process must be highly optimized. This paper describes the results obtained by a distributed group of students working on this problem in an open source manner.

Introduction

Description of schisto, PZQ and the need for (R)-PZQ
Description of the existing process
Identification of current shortcomings
Brief outline of how the project was intended to operate

Results and Discussion

Chemistry

The first step towards the enantioselective isolation of R-(–)-Praziquantel (PZQ) from racemic PZQ necessitates the acid-promoted hydrolysis of PZQ to racemic praziquanamine (PZQamine), as shown in the diagram below (Figure 1):¹

Figure 1: Acid Promoted Hydrolysis of PZQ

The presence of hydrochloric acid is necessary in order to activate the carbonyl groups of the amides of PZQ, thereby resulting in the formation of an electrophilic carbon which is significantly more susceptible to nucleophilic attack.^1,2 Nucleophilic attack by either of the weak nucleophiles present in solution, namely water or ethanol, is therefore able to occur.^1,2 Indeed, the hydrolysis proceeds mechanistically in the following manner, yielding the desired PZQamine, a carboxylic acid, as well as a side product ester (Figure 2):^1,2

Figure 2: PZQamine Hydrolysis Mechanism

It must be noted that hydrolysis of the amide bonds involved is a reversible process.² However, the relative abundance of water and ethanol leads to the promotion of the hydrolyzed form of PZQ.² Interestingly, the second amide bond present in PZQ is prevalently not hydrolyzed due to the kinetic favorability of intramolecular reactions, thereby accounting for the largely selective hydrolysis of the PZQ to yield the desired intermediate PZQamine.² The greater steric bulk of the unhydrolyzed amide bond further contributes to the decreased nucleophilic attacks of water and ethanol and, consequently, to the overall preservation of the unhydrolyzed amide bond.²

Extensive spectral data collected confirm that the desired PZQamine intermediate was successfully synthesized (1.7883 g; 69%; under 1M HCl). Moreover, the spectral data extensively contributes to verifying the validity of the previously discussed proposed hydrolysis mechanism and providing insight regarding the optimization of HCl concentrations employed in PZQ hydrolysis. The IR spectrum (KBr pellet) of the isolated product of PZQ hydrolyzed under 1M HCl conditions is wholly consistent with the expected results. The appearance of a strong, broad peak at 3314.39 cm^-1 corresponds to the presence of N–H stretching. Given that such a peak, in this experiment, would only be observed in the event of the successful hydrolysis of PZQ to form the secondary PZQamine, the IR results provide convincing evidence that the desired reaction did indeed proceed. Furthermore, the presence of a peak at 1636.04 cm^-1 is consistent with the stretching of an amide C=O. Therefore, such a peak suggests that the product does indeed contain an amide group, which is consistent with the stipulation that the second amide of PZQ will not be hydrolyzed. While such a peak does not in itself exclude the possibility of the presence of residual amounts of starting PZQ as well, TLC results (Silica; 1:1 Hexanes : EtOAc; CAM Visualization) are indicative that the starting material was completely consumed in the course of the reaction. As such, the TLC results provide strong evidence that the amide C=O stretch observed is primarily due to the PZQamine intermediate.

The ¹H NMR spectrum of the isolated product is in accordance with NMR results reported in literature, with the exception of two peaks, namely those at 4.1 ppm and 1.2 ppm, respectively.¹ Indeed, the similarity of the NMR obtained to that of PZQamine is a strong indicator that it is highly likely that the desired hydrolysis was achieved. Furthermore, the two peaks which are incompatible with literature NMR values are consistent with peaks for either the expected side product ester (ethyl cyclohexanecarboxylate) or residual ethanol. A specific determination regarding which of these are present is not possible, given that peaks that would distinguish the ester from ethanol coincide with peaks corresponding to the product PZQamine. Nevertheless, the NMR confirms that the desired intermediate was indeed synthesized, albeit with impurities present. It is therefore clear that the purity of the synthesized PZQamine is less than 100%. Consequently, impurities contribute to the measured yield of 69%, suggesting that the actual yield of PZQamine is less than 69%. Given that, under similar reaction conditions, a yield as high as 92% is reported in literature, it is clear that there is a considerable discrepancy between observed and expected yield.¹ Potential explanations for yield inconsistency include product loss in transfers as well as the implementation of fewer extractions aimed at product isolation than suggested in literature.¹

Based on preliminary experimentation regarding the effects of varying HCl concentrations on PZQ hydrolysis, no noticeable difference was observed between results obtained using 1M and 2M HCl in terms of IR, TLC, and NMR results. Furthermore, while the IR and NMR results of hydrolysis under 0.5M HCl largely matches those obtained under 1M and 2M HCl, the presence of two distinct spots on the TLC result indicates the substantial presence of impurities or unreacted PZQ. Due to initial unfamiliarity with the reaction, the yield data was determined before all of the solvent was successfully evaporated, and is therefore not of value in the comparison of the HCl concentration influence on PZQ hydrolysis. In essence, this aspect of the experiment would have to be repeated before any conclusive claims can be made concerning the optimization of HCl concentration in PZQ hydrolysis.

In the future, optimization of the hydrolysis of PZQ to PZQamine could be pursued through further, more detailed, experimentation on the effects of HCl concentration on PZQ hydrolysis. Furthermore, the implementation of more extractions followed by purification by recrystallization would perhaps contribute to increasing both yield and purity of the hydrolysis product. Moreover, characterization with the aid of GC-MS and ¹³C NMR would perhaps contribute to the more accurate identification of the postulated side products. Having successfully obtained the intermediate racemic PZQamine, the next step of the experiment involved the implementation of L-tartaric acid derivatives as resolving agents for the isolation of R-(–)-PZQamine from the racemic PZQamine (Figure 3):¹

Figure 3: PZQamine Resolution

Resolution was achieved through the formation of a salt between the conjugate acid of PZQ amine and the conjugate base of the L-tartaric acid derivatives whose stereochemical configurations were mutually compatible, as shown below (Figure 4):^1,2

Figure 4: PZQamine Resolution

In this case, the presence of the L-tartaric acid derivatives in a comparatively basic environment allowed for the formation of the conjugate base of the L-tartaric acid derivatives which promote resolution.^1,2 Furthermore, the resulting acidified mixture allowed for PZQamine, the most basic species present in the reaction, to be protonated to form its conjugate acid, as seen in Figure 4.^1,2 The acid-base reactions described contribute to the success of the resolution, given that the formation of a salt between the resolving agent and a chiral compound is aided by the resulting electrostatic interactions.²

While resolution of PZQamine had previously been achieved through the implementation of a number of L-tartaric acid derivatives, the two particular resolving agents applied in this experiment, namely (−)-O,O′-Di-pivaloyl-L-tartaric acid (3) and (−)-O,O′-Di-p-toluoyl-L-tartaric acid (4), had not been utilized or tested in the isolation of a single enantiomer of PZQamine.1 For reasons which remain unclear, (3) did not result in the formation of a salt/precipitate when subjected racemic PZQamine in the presence of isopropanol and water, despite its structural similarity to successfully implemented resolving agents (Figure 3).¹ Further experimentation notwithstanding, the experimental results strongly, if not conclusively, indicate that (3) unfortunately cannot be utilized as a resolving agent in the resolution of PZQamine. Based on spectral data collected, (4) was successful in the resolution of PZQamine to yield R-(–)-PZQamine (5) (0.1110 g; 44%; 79% ee). Given that the resolution of PZQamine does not yield a product which has different IR or ¹H NMR spectra from the starting material, the IR and ¹H NMR results only serve to confirm that the product is the desired PZQamine. Most importantly, the IR spectrum (Thin Film) of the resolved PZQamine reveals, among others, a peak at 3367.36 cm^-1, consistent with the presence of N–H stretching, as well as a peak at 1636.16 cm^-1 corresponding to a C=O amide stretch. Indeed, the IR spectrum of the resolved PZQamine is wholly consistent with that of racemic PZQamine, as expected. Furthermore, the ¹H NMR spectrum of the resolved intermediate is in complete agreement with NMR results reported in literature, with the exception of peaks at 4.0 ppm and 1.2 ppm, respectively.¹ The two incompatible peaks mentioned are consistent with peaks for isopropanol. As such, the isolated compound is not 100% pure. It follows from such a conclusion that the calculated 44% yield is in fact higher than the actual yield. Nevertheless, the yield value of 44% is consistent with literature values available for similar resolving agents implemented in PZQamine resolution, and is not far away from the optimal figure of 50%.¹ It should be noted that, although performing the experiment a second time resulted in a yield of 118% (0.8275 g), such a yield was obtained due to inadequate solvent evaporation. As such, the figure of 44% used in the above discussion is far more realistic and representative of the reaction. Repeating the experiment and diligently drying the isolated substance would provide valuable information regarding the yield of resolved PZQamine achieved by (4). Furthermore, given that ¹H NMR data strongly suggests that isopropanol is the only impurity present in a detectable quantity, it is not unreasonable to conclude that non-trivial quantities of side products were not synthesized. The ¹H NMR spectrum of the product with a shift reagent, in this case Eu(hfc)₃, provides the information necessary to determine the enantiomeric excess of the PZQamine, and thereby judge the effectiveness of (4) as a resolving agent. Using the peaks at 4.33 ppm and 4.7 ppm, which correspond to the chiral hydrogen of PZQamine, the enantiomeric excess achieved was determined to by 79%. As such, the resolving agent (4) contributed to the fairly successful enantioselective isolation of the desired R-(–)-PZQamine. Indeed, the enantiomeric excess achieved is comparable to that attained by other previously implemented resolving agents, such as (–)-dibenzoyl-L-tartaric acid.¹ Overall, the resolution of PZQamine using (4) to yield R-(–)-PZQamine (5) (0.1110 g; 44%; 79% ee) can be declared to have been highly successful. Future endeavors aimed at the resolution of PZQamine should perhaps center upon the exploration and development of new resolving agents, both those derived from L-tartaric acid and those from perhaps less conventional resolving agents. Furthermore, subsequent undertakings of this synthesis should likely include further purification steps, most likely by recrystallization, in the interest of increasing product purity. Having attained the successful enantioenrichment of PZQamine, the final step towards the synthesis and isolation of R-(–)-PZQ (6) entails adding cyclohexanoyl chloride to (5), as shown in the diagram below (Figure 5):¹

Figure 5: Synthesis of R-(–)-Praziquantel

The electrophilic nature of the chlorine atom of the acid chloride results in the presence of an electrophilic carbon which is highly susceptible to nucleophilic attack.² As such, the secondary amine of the PZQamine will perform a nucleophilic attack on the acid chloride, leading to the synthesis of PZQ, as shown mechanistically below (Figure 6):²

Figure 6: Mechanism of the Synthesis of R-(–)-Praziquantel

Interestingly, given that it has been established above that the resolved PZQamine (5) contains the comparatively more nucleophilic isopropanol as a substantial impurity, a side product ester, isopropyl cyclohexanecarboxylate, would result from the reaction of the residual isopropanol and cyclohexanoyl chloride, as shown below (Figure 7):

Figure 7: Mechanism of the Side-Product Isopropyl Cyclohexanecarboxylate Synthesis

As the isopropanol impurity consumes a non-trivial quantity of the acid chloride reactant, it logically follows that the precursor resolved PZQamine will itself be an impurity in the final product.

The spectral data accumulated convincingly indicates that the desired synthesis of the target R-(–)-PZQ (6) (0.3846 g; 83%; 90% ee) did indeed occur. Among the many peaks produced, the IR (Thin Film) spectrum of the isolated powder reveals a significantly smaller peak corresponding to N–H stretching (3490.75 cm-1) than was observed in the IR spectrum of (5). It is reasonable to conclude that, while still present in the final product, the PZQamine starting material (5) was consumed by the reaction to a considerable extent. Nevertheless, the presence of the N–H stretch peak convincingly implies that the final step of the synthesis did not go to completion. The proposed explanation for such an occurrence, namely the reaction of residual isopropanol with the cyclohexanoyl chloride to form an ester side product, is itself given credence by the presence of a peak at 1722.49 cm-1, which is consistent with the C=O stretching of esters. Therefore, the IR spectrum provides strong evidence that the desired reaction occurred to a significant extent, in addition verifying the presence of the postulated side-product ester. As such, the IR spectrum lends significant credibility to the mechanism of the synthesis of (6) proposed above. Furthermore, as the IR spectrum identifies both unreacted PZQamine (5) and the side product ester as impurities present in the product powder, it follows that the true yield of (6) is less than the calculated 83%. Nevertheless, the product yield of the final step of 83% is largely consistent with the yield values reported in literature (90%).¹

The ¹H NMR spectrum of the product, while not optimal due to excessive dilution of the product leading to an inadequate and unclear spectrum, is in agreement with the expected NMR of PZQ. In essence, while the NMR spectrum does not contradict the above interpretation of the IR spectrum, it is not suitable for any detailed analysis or confirmation of potential side products. It would therefore be of interest to produce a more adequate NMR spectrum of the isolated product in order to potentially pursue a more rigorous product analysis. Moreover, the ¹H NMR spectrum of the product with the shift reagent Eu(hfc)₃ unfortunately does not lend itself at all to any analysis of the enantiomeric excess achieved.

Perhaps most importantly, the polarimetry measurement of the final product (λ = 589 nm; c = 1.039 g/(100 mL); DCM) of –127° corresponds to a calculated enantiomeric excess of 90% based on polarimetry values of enantiopure R-(–)-PZQ available in literature.¹ As such, despite the inadequacy of the shift NMR in determining enantiomeric excess, the polarimetry data is a clear and unambiguous indicator that the synthesis undertaken was indeed extensively enantioselective. Interestingly, the calculated enantiomeric excess value of 90% exceeds the measured enantiomeric excess of the precursor substance (79%). Given that resolution had occurred in the synthesis of the precursor (5), it is not possible that the enantiomeric excess of the subsequent product is greater. Interestingly, the fact that the measured enantiomeric excess value of the final product (6) is unrealistic can be accounted for by the presence of unreacted resolved PZQamine (5) in the product powder. As the specific rotation of enantiopure R-(–)-PZQamine is more negative (–152°) than that of R-(–)-PZQ (–140°) based on values available in literature, the presence of R-(–)-PZQamine as an impurity in the final product would have resulted in a more negative polarimetry measurement, which in turn translates to a greater calculated enantiomeric excess value.¹ It must be noted that, while the desired final product (6) and precursor (5) were enantioselectively synthesized, neither are enantiopure. As such, while the synthesis was successful in the isolation of enantioenriched R-(–)-PZQ, the evidence discussed above indicates that the goal of synthesizing enantiopure R-(–)-PZQ was not achieved.

Given the presence of substantial impurities, future studies implementing the above synthesis of (6) should include recrystallization steps in the interest of increasing purity. Furthermore, future experiments should utilize GC-MS and ¹³C NMR in the interest of more precise identification of the side products present in the isolated substance.

Conclusion

The hydrolysis product of racemic praziquantel (PZQ), namely racemic praziquanamine (PZQamine) (1.7883 g; 69%; under 1M HCl) was successfully resolved using (−)-O,O′-Di-p-toluoyl-L-tartaric acid to enantioselectively isolate R-(–)-PZQamine (0.1110 g; 44%; 79% ee), which was itself then implemented in the successful enantioselective synthesis of R-(–)-PZQ (0.3846 g; 83%; 90% ee). Although the synthesis undertaken achieved the isolation of enantioenriched R-(–)-PZQ, evidence suggests that the goal of synthesizing enantiopure R-(–)-PZQ was not attained. Overall, this work illustrates the feasibility of implementing simple and comparatively inexpensive laboratory techniques and reactions towards the enantioselective synthesis of select medicinally-relevant compounds.

References

(Please include DOIs to any references provided)