Resolution of Praziquanamine

Published by MatTodd on 15 April 2010 - 8:43pm

Subject 

Request for Help

PZQ can't be resolved as-is (unless anyone has any bright ideas how to resolve amides). One of the most promising strategies to prepare enantiopure praziquantel (using a strategy that starts from the racemate) is a classical resolution of praziquanamine (1, "PZQamine"). This molecule can either be made from scratch (it's an intermediate in the current PZQ synthesis) or can be obtained in high yield from PZQ itself.
 
 

 
PZQamine a secondary amine. We need a resolution that is inexpensive, and will work on a large scale.
One of the original patents in this area (US 3,993,760, 1976) details a way to do this resolution, shown below. This patent actually concerns a benzoyl analog of PZQ, but the effect is the same since PZQamine is the amine resolved.
 

 
Quinic acid is employed. The initially-formed crystals are the salt formed from the undesired enantiomer of PZQamine, so the mother liquor is concentrated to yield new crystals of the desired enantiomer. The exact procedure as reported is (page 5, column 7, "Example 1" lines 32-50):
 
"[(-)-PZQamine] may be obtained as follows: 24.3 g. of [(rac)-PZQamine] (m.p. 118-119 oC; obtainable by treating [(rac)-benzoyl analog of PZQ] with methanolic hydrochloric acid and subsequent heating to 180 oC at 15 mm Hg) are dissolved in 100 mL methanol and added to a warm solution of 30 g. quinic acid in 500 mL methanol. The mixture is refluxed for 15 minutes and then cooled to 20 oC. The crystals obtained are separated by filtration, the mother-liquor is evaporated to 100 mL and the crystals precipitating now are separated. The quinic acid salt of [(-)-PZQamine] is obtained; m.p. = 196 oC. The salt is dissolved in water, the solution is made alkaline and extracted with chloroform. After drying and evaporating the organic solvent the laevorotatory free base is obtained; m.p. = 120 oC, [a]D 20 = -306 o C (sic)"
 
The most obvious thing about this procedure is that the mass recovery of the desired enantiomer is not given. It is also doubtful that enantiopure material is produced without a final recrystallisation, but this is not specified. Nor is the solvent/concentration employed in the measurement of optical activity. Having to obtain the desired material from the mother liquor is also sub-optimal. This is a problem with using quinic acid as a resolving agent, since the unnatural enantiomer is not available in quantity.
 
Update August 27th 2010: Syncom BV rapidly identified a simple resolution method for PZQamine below. This method employs either quinic acid (which was also mentioned in the original Merck patent on PZQ) or dianisoyltartaric acid. (-)-PZQ is the active enantiomer desired in this project. Resolution with naturally-occurring D-(-)-quinic acid gives (+)-PZQamine. If we make the assumption that (-)-PZQamine leads to (-)-PZQ, and if we assume that we want to isolate PZQ from the solid component of the resolution rather than the mother liquor, then we would need L-(+)-quinic acid for this PZQ resolution which is not available. However (-)-dianisoyl-L-tartaric acid also gives (+)-PZQamine, implying (+)-dianisoyl-D-tartaric acid will give the desired (-)-PZQamine and hence (-)-PZQ. Both enantiomers of dianisoyltartaric acid are available commercially (for the same price), e.g. from TCI. The procedure and spectra are shown below, and will now be repeated in our lab here in Sydney. We'd encourage others to verify/check this also.
 
Syncom have done a fantastic job here. If there are remaining questions, they are these:

1)   Which other chiral acids are good candidates for this resolution?
2)   Which other solvents should we examined?
3)   Which conditions should we optimize to keep the eventual cost of the large-scale process low?
4)   Are there any good guidelines for procedures to maximize the yield of crystalline salt, and the yield of the subsequent re-isolation of the free amine?
5)   Does anyone know of papers where other amines resembling praziquanamine (with a stereocentre a couple of carbons away from the NH) have been resolved?
6)   What can we do with the undesired (S)-enantiomer of praziquanamine? Throw it away? Our default plan is to dehydrogenate and re-hydrogenate (via 2, above) to re-form (rac)-praziquanamine, but this is not elegant.

Relevant publications and groups (please just edit the page and add below using DOI to link):
Screen for resolution conditions using composition of mother liquor (Eli Lilly)
Merck Patent
 
Please feel free to post any other suggestions below. Even better – plan to have a go in your own lab and post data here.

 

Comments

Michael Wolfle's picture

We started with attempts to the racemic resolution of praziquanamin.Various chiral carboxylic acids have been tested for resolving the racemic mixture of 1:

L-(-)-malic acid
L-(+)-tartaric acid
(-)-2,3-dibenzoyl-L-tartaric acid
S-(+)-mandelic acid
(1S)-(+)-camphor-10-sulfonic acid
(1R,3S)-(+)-camphoric acid
(+/-)-BINOL

The standard solvents usually used for a resolution of diastereomeric salts are polar solventsliquids like H2O and light alcohols such as MeOH and EtOH. I have varied the solvents and the combination of the following: H2O, MeOH, EtOH, iPrOH, acetone, acetonitrile, EtOAc, Et2O, chloroform, toluene.
The major problem is the crystal formation of the diastereomeric salts – almost all of the mixtures were obtained as oils and some of them as amorphous solids. Only in a few samples I achieved a crystal precipitation, but these consist of a nearly racemic mixture of praziquanamin (observed by optical rotation).

Crystal forming mixtures: (Resolving agent (ratio resolving agent:amine), solvents)
(1S)-(+)-camphor-10-sulfonic acid (1:1), Et2O, Et2O/acetone
L-(+)-tartaric acid (1:1), EtOH/H2O, iPrOH/H2O, Acetone/H2O

Literature to racemic resolution:
[1]    “CRC Handbook of the Optical Resolution via Diastereometric Salt Formation”, D. Kozma, CRC press, 2002.

[2]    “Optical resolution methods”,  E. Fogassy, M. Nógrádi, D. Kozma, G. Egri, E. Pálovics and V. Kiss, Org. Biomol. Chem. 2006, 4, 3011-3030; DOI: 10.1039/b603058k.

[3]    “Industrial Methods for the Production of Optically Active Intermediates”, M. Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Keßeler, R. Stürmer and T. Zelinski, Angew. Chem Int. Ed. 2004, 43, 788-824; DOI: 10.1002/anie.200300599.
 

ndt228's picture

Dear Matt,
Here are some thoughts arising from your posting:
1) Concerning your screen for classical resolving agents - past experience would suggest that you need more diversity with your chiral acids, and less immediate attention to the solvents, until you get hits. Its a good rule of thumb to stick with EtOH-water mixtures (choose one offering good average solubilty for the amine & the acid). There are 30+ reasonably available chiral acids, and no good way to pre-select one over another. When you are looking for hits don't be too concerned about long-term cost.
2) If you haven't done this already, use chiral chromatography to generate some single enantiomer material. Use this to study the properties of the single isomers - do they crystallise as free base, or as (eg) HCl salt ? Determine solubilty data across the range from 0-100% ee, plot the phase diagram & use this to predict the operating window for your resolution process. Further, you can input single enantiomer amine into a salt-screen of potential resolving agents to see whether any of them will crystallise under the optimum conditions. If you get any hits you can feed this back into the hunt for a salt resolution.
3) Optical rotation alone is an insensitive measure of quality (in general); much better to get (analytical) chiral hplc working on your amine.
4) A zero rotation or 1:1 peak ratio does not mean a resolution candidate has failed. If both salt forms are of low solubility under the conditions trialled, this will give a low ee outcome. You need to consider the yield as well. The max theory yield is 50% (unless you manage to get dynamic kinetic resolution to work). Hence if you get a 75% yield of "salt" the maximum chiral purity it could have would be 67% ee.
So, in your screening, do not reject "low" yields - and check the ee of both the product and the liquors. Chiral lc will help here as it can be much higher throughput than OR, and needs a lot less sample.
5) A zero rotation does not rule out a "conglomerate" salt - not that I am suggesting that you should hand-pick your crystal crop like Pasteur, but there are different techniques needed to exploit a conglomerate resolution. If there is a "tame" crystalloographer in your Uni, and you have nice crystals (he/she will tell you what they consider workable - often it's not what a synthetic chemist would pick), then it would be fairly quick to establish a crystal structure - this will define absolutely whether it is a racemic crystal or not. NB Sometimes even achiral acids can give conglomerate salts ! 
best regards
Nick

MatTodd's picture

Nick - great advice, thanks.
1) We've been screening some other chiral acids - Michael will post on this soon. Then we'll look into solvent variations.
2) Obtaining a few grams of pure enantiomers of PZQ or the praziquanamine would be very useful - we have no way of doing this with chiral LC locally - if anyone reading would like to volunteer this support, that'd be enormously appreciated and we could mail the raw material.
3) Plotting the phase diagram would be a great thing to do - unless this is already known?
4) I agree that in search mode we need to employ chiral LC of crystals and mother liquor. I'm a big fan of optical rotation in general (as my group knows) but we don't want a false negative here.
5) Yes, yields are important and we need to report those also.
6) I suspect the crystal structure is already known... but I don't have it to hand.
 

ndt228's picture

Another thought occurred after my first post - forgive me for asking but when you said that all of your crystalline salts had given near-zero optical rotation:
a) Did you mean the salts themselves, or the amine released from them  ?
b) Do you have any idea what size of optical rotation the pure amine enantiomer has ?
because:
1) If it was the salts that were tested then even if the amine component was racemic (ie zero rotation) then the chiral acid should still have put a twist into the light ie giving a non-zero OR.
2) I have never run OR on chiral salts - it seems theoretically possible that by chance the rotation of chiral amine in one direction could be cancelled out by the opposite twist from the chiral acid.  Of course pretty unlikely that this would happen with two different salts.....but it's one reason I much prefer chiral chromatography (though that too can have interferences, and you must routinely run racemic stds to ensure you maintain resolution)
3)  Optical rotations of my (limited) experience can vary from tiny to huge - one I had was ca 550 degrees - and precision of measurement is not that great for defining high-end chiral quality.  Again, why I like chiral lc.
 
On which subject I have done a little searching for chiral methods.  You may know of these already, but I attach three methods for praziquantel itself.  These are from one well-known supplier of Chiral columns - there are others out there......
 

ndt228's picture

I found this: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2614124/
and wanted to be sure you knew about it. Key content includes: (my bold emphasis)

Preparation and stereochemical assignment of (−)-PZQ
Although effective synthetic methods for the enantioselective preparation of PZQ have been reported [20], we opted for the direct enantioseparation of the racemate yielding gram quantities of both optical forms. The preparative scale chromatography was performed on microcrystalline cellulose triacetate using methanol as the mobile phase, conditions under which the enantiomer having the negative optical rotation emerged first from the column [21]. After crystallisation from methanol/water, (−)-PZQ was obtained in enantiomeric excess >99%, as determined by HPLC (column used Chiralcel OD-H). No residual other enantiomer (+)-PZQ was detected in this sample. X-ray structural analysis, using Cu-Kα radiation, of a monoclinic crystal in hemi-hydrate form obtained from said fraction by crystallization from methanol/water unequivocally proved the R-configuration of the molecule by measuring Friedel pairs and the Flack parameter (x = −0.1(3)) (Figure 2Figure 2). Further details of the crystal structure analysis are available on request from the CCDC (www.ccdc.cam.ac.uk) quoting the names of the authors and journal citation.

So, you are now equipped with a practical way to get to gram qtys of (+)- and (-) PZQ, from which you can obtain the amine by your established hydrolysis.  Then you can make a screen for crystalline salts & hopefully work back from there to a resolution of the amine.  Then all you have to to is make the amine !
all the best, Nick

MatTodd's picture

Nick - thanks, yes the bitter taste of PZQ is a big reason to generate the enantiopure material - if we can remove the (S) enantiomer we can remove the bitterness, which will greatly increase compliance in the millions of people taking the drug. I'd forgotten that they used that method to obtain their material. We've no experience in doing this technique/derivatizing the cellulose, so even though it sounds simple we may ask Intervet for some help here. Great you spotted this, Nick, thanks.

ndt228's picture

Matt - I have already written to Christian (of the "Taste" paper) asking for details of the separation they used.  Meanwhile other web-searching has shown that the stationary phase is reasonably cheap (as such things go) and can even be bought from Aldrich (though at a different particle size; I have not yet found a price from Merck (Darmstadt).  Also, microcrystalline Cellulose triacetate has been quite widely used for resolutions, since the 1990's.  If you'll leave it with me for a few days I'll report back when I have more information. 
Meanwhile, what & where are the facilities at Intervet ?  I think they are part of Schering Plough, who have recently been swallowed by Merck & Co (prev = Merck, Sharp & Dohme, MSD) From personal experience there may be quite a lot of turmoil within the entire group of companies, whilst management seek out "synergies".  This is certainly the on-going story within Pfizer, GSK & Astra-Zeneca.
Cheers, Nick

MatTodd's picture

This page is getting pretty long - we need to assimilate the experimental things that need doing into the start of the page, and continue to update that as results come in so there is a clear plan of attack. This is the problem with blog-type pages...
Some of the content here (e.g. the great advice from Nick) arose from a question I posted on LinkedIn, which has a fair number of process chemists on it. Sadly it's a closed discussion area, which is not in the spirit of Open Science (lower those walls, guys!). However we also made contact with the corresponding author of the Lilly paper linked further up this page, now at Solvay, Alfio Borghese. He helpfully supplied this advice for the praziquanamine resolution:
 
"Two data are important to assess the efficiency of the resolution:

  • The ratio of the diastereomers in the mother liquors (obtained by chiral HPLC). This in fact will give you the eutectic composition.
  • Then to optimize the process conditions, you need the solubility of the eutectic (obtained by evaporation of the ML and weigh it.)

The best way to generate these data is to repeat the crystallization experiment in an as concentrated as possible manner  in the solvents already used. Let the reaction mixture stabilize for at least 24h to reach the thermodynamic equilibrium. Then filter it and do the analysis of the mother liquors by chiral HPLC in order to get the ratio between the two diastereoisomers. With these values I can asses the efficiency of the resolution."
 
Thank you, Alfio.

Michael Wolfle's picture

Nick,
Thanks for your valuable advice.
1)    Indeed, a higher diversity of the resolving agents is necessary. First I’d like to test all the chiral acids we have in stock (it is also a cost factor). Therefore I’m very pleased that we can get help from the Syncom BV in the Netherlands.
2)    That’s a thing I should consider. PZQamine can be crystallized as free base, but I haven’t obtained enantiopure material yet. Chiral chromatography with works well with PZQ on cellulose triacetat, but I’m not sure if it’s possible with praziquanamine (see 3).
3)    Optical rotation is for the moment the only method I can use for the determination of the ee of PZQamine. Trials with chiral HPLC have failed yet, when I was using a Chiralcel OD-H column (cellulose derivative as stationary phase) (hexane:IPA:TEA = 80:20:0.1, RT ~ 20 min, broad peak of both enantiomeres).
4) In the previous experiments the yield of the precipitated salt was also considered. If I got more then 50% precipitate the mother liquor was diluted.
5) Is there a rule of thumb, how the crystals should look like? In most of the cases I got an amorphous precipitate (but still an ee). Of course, it would be nice to have a crystallographer which could supply me, but I don’t think that I could convince him/her for helping me with my screening experiments…
Thanks,
Michael.
 

ndt228's picture

Dear Michael,
Re chiral hplc analysis of PZQAmine samples - see the recent posting now under the PZQAmine Resolution header which details results at Syncom.
Regarding suitable crystals for XRD work:  a) Smaller than you'd think (generally best viewed at 10-20x magnification  b) transparent, "perfectly formed" - ie a single crystal, not a clump of them growing in different directions  c) preferrably (but usually outside your control) a chunky shape rather than whispy needles or thin plates  d) kept in contact with Mother Liqs until delivered to crystallographer - just in case it is a solvate (can fall apart if the solvent is pulled out on vac pump or even in open air). 
Within these general targets I usually take 100 mg or so of crystals and let the crystallographer take his/her pick of them.  They have a variety of techniques for mounting them for data collection and may try 2 or 3 times to get one that gives good results.
cheers
Nick

Has anyone suggested trying to develop a protein that selectively binds one of the enantiomers (of either PZQ or an intermediate)? There are several possibilities:
1. Search for a bacterium that can live on the undesirable enantiomer. (A good place to look is downstream of a PZQ plant.) Such a bacterium should produce an enzyme that degrades the wrong enantiomer.
2. Screen a schistosome cDNA expression library for PZQ binding. If PZQ has a protein target (it might not) it would probably show up & should bind the right enantiomer only.
3. Use combinatorial chemistry/mutagenesis to evolve a PZQ-binding protein.
4. Design a PZQ-binding protein. (A DDT-binding protein was successfully designed way back in 1983 - FEBS Letters 157 247.)
5. Do a mass screening of as many proteins as you can.

Oops, I left out the most straightforward possibility:
6. Raise antibodies against PZQ.

MatTodd's picture

I have no expertise in these areas - maybe someone else can chime in. My gut tells me they're going to be expensive..?

  • Screening of a number of chiral acids at Syncom B.V. revealed that both quinic acid (as was reported in a Merck patent US3993760A1 example 1;  1976) and (-)-dianisoyl-L-tartaric acid displayed excellent resolving abilities for racemic praziquanamine, kindly provided by Matthew Todd. One recrystallization afforded the salt with a de of 95%, enriched in the (+)-enantiomer. The exact yield has not been determined but with some optimization it should be possible to obtain a yield of at least 30% (60% of the theory).
  • Non-optimized Syncom protocol for resolution of praziquanamine with (-)-dianisoyl-L-tartaric acid:
    A mixture of rac-praziquanamine (101 mg, 0.5 mmol) and (-)-dianisoyl-L-tartaric acid (214 mg, 0.5 mmol) was dissolved in a mixture of isopropanol (2 mL) and water (0.4 mL) by the application of heat. The obtained clear solution was allowed to cool to room temperature overnight and the resulting crystals were isolated by filtration. The free base was liberated with aqueous base, extracted and analyzed by chiral HPLC using UV detection (220 and 260 nm) and a PDR chiral advanced laser detector. Various chiral HPLC methods were developed at Syncom B.V. and were posted on the website http://www.thesynapticleap.org/node/315. An ee of 65-67% was found for the (+)-isomer, using a chiral advanced laser detector at 675 nm. The filtrate (mother liquor) was enriched in the desired opposite (-)-enantiomer (ca. 60% ee). One recrystallization of the crystals from a mixture of isopropanol and water (2 and 1 mL) increased the ee to 95%. (Dr. Kees Pouwer, Syncom B.V., The Netherlands, May 26, 2010).
  • The diastereomeric salt resolution of praziquanamine with diaroyl tartaric acids is simple and can easily be upscaled to produce kilogram quantities or more. For the preparation of (-)-praziquanamine the resolution with (+)-dianisoyl-D-tartaric acid is preferred. When using the (-)-dianisoyl-L-tartaric acid the (-)-praziquanamine enantiomer should be isolated from the mother liquor.

Jean-Paul
(Note this information is also posted here)