Frakefamide, an analgesic tetrapeptide: Development of a pilot-plant-scale process

Article abstract of DOI:10.1021/op020060k

A pilot-plant-process is described where frakefamide × HCl (L-tyrosyl-D-alanyl-p-fluoro-L-phenylalanyl-L-phenylalanin-amide hydrochloride) was synthesised from its amino acid monomers in seven steps. The synthesis was performed in 70-L equipment, and the

Full text of DOI:10.1021/op020060k

Organic Process Research & Development 2002, 6, 788−797
Frakefamide, an Analgesic Tetrapeptide: Development of a Pilot-Plant-Scale
Process
Henry M. Franze´n,† Galina Bessidskaia,‡ Vahak Abedi, Anders Nilsson,§ Maths Nilsson, and Lars Olsson*
Process R&D, AstraZeneca R&D So¨derta¨lje, S-151 85 So¨derta¨lje, Sweden
Abstract:
A pilot-plant-process is described where frakefamide × HCl
(L-tyrosyl-D-alanyl-p-fluoro-L-phenylalanyl-L-phenylalanin-
amide hydrochloride) was synthesised from its amino acid
monomers in seven steps. The synthesis was performed in 70-L
equipment, and the final product was obtained in 70% overall
yield and in 99.5% purity. Only two intermediates were isolated,
and the process required no chromatography. Peptide bond
formation was promoted by isobutyl chloroformate-mediated
mixed anhydride coupling reactions. The formed mixed anhy-
drides proved to be surprisingly stable, in most cases for several
hours at -10 °C, and therefore suitable for large-scale peptide
synthesis. Only traces, if any, of racemised coupling products
were obtained. Benzyloxycarbonyl was used as amino protecting
group throughout the synthesis, and its removal by hydro-
genolysis proved to be fast and convenient on a large scale.
Figure 1. Frakefamide (1a).
cleavage have been suggested as reasons for the better
activity.3 For drug development the hydrochloride salt,
frakefamide × HCl (1b), would be more suitable as drug
substance rather than the neutral peptide. Any efforts for
large-scale production would therefore be directed towards
this compound.
For pre-clinical studies, investigational product in amounts
up to 100 g was provided by solid-phase synthesis. A clinical
development program would require multikilogram amounts,
and future commercial production for market needs could
be predicted in the range of tonne quantities per annum. A
solid-phase approach was not considered realistic for this,
mainly due to suspected problems with capacity. Costs would
also be a problem for a solid-phase method even if enough
capacity could be reached. The need for a large-scale solution
synthesis was obvious, and the results from the development
of an advanced pilot-plant process are described below.
Introduction
Frakefamide (1a) is a tetrapeptide with an amide function
at the C-terminus (Figure 1). The amino acid sequence of
frakefamide is L-tyrosyl-D-alanyl-p-fluoro-L-phenylalanyl-L-
phenylalaninamide. Hence, it contains both natural and
unnatural R-amino acids. Frakefamide (1a) has shown
interesting properties for use in acute and chronic pain
treatment.1 It acts as a selective agonist on µ-opioid receptors2
and belongs to the dermorphin peptide family.3 These
peptides show analgesic properties and were discovered in
the early 1980s. Dermorphin itself (Tyr-D-Ala-Phe-Gly-Tyr-
Pro-Ser-NH2) and other naturally occurring peptides in the
dermorphin family have a conserved N-terminal sequence
of Tyr-D-Ala-Phe. The D-Ala residue is essential for activity;
corresponding L-Ala peptides are practically inactive.4 Fur-
thermore, the amide function at the C-terminus is also
important for activity; it increases the affinity to the µ-opioid
receptor about 30-100 times compared to carboxyl group
analogues. Suppression of the negative charge from a car-
boxyl group and better protection against carboxypeptidase
Results and Discussion
Both a linear and a convergent solution synthesis of
frakefamide × HCl (1b) can easily be designed. For a peptide
of this size it is not obvious that one approach would be
superior to the other. The number of synthetic steps is similar
so other factors are important. These include crystallinity of
intermediates, possibility to outsource the synthesis of
intermediates, design of process logistics, and the obtained
quality and yield of the final substance. From early on,
energy was devoted to a linear synthesis that was utilised in
the pilot plant process presented here. However, a convergent
block synthesis was also developed to challenge the linear
route. For both approaches there was initially a need to define
some basic chemistry, a peptide coupling method, and a
protecting group strategy suitable for large scale.
Coupling Methodology. Ideally, all peptide bonds in
frakefamide (1a) should be formed by identical reactions to
allow an efficient process flow with the best operability.
Finding a suitable coupling reagent was therefore a key issue,
and many aspects had to be considered and examined. The
reagent should first of all be efficient (high-yielding, fast)
and possible to handle on a large scale (relatively inert,
* Corresponding author. E-mail: lars.rg.olsson@astrazeneca.com. Tele-
phone: +46-8-553-230-48.
‡ Present address: Medicinal Chemistry, AstraZeneca R&D So¨derta¨lje, S-151
85 So¨derta¨lje, Sweden.
§ Present address: Syntagon AB, Box 2073, S-151 02 So¨derta¨lje, Sweden.
† Present address: Medicarb AB, Annedalsva¨gen 37, S-168 65 Bromma,
Sweden.
(1) Preclinical development has been made in collaboration with Biochem
Pharma, Montreal, Canada, now part of Shire Pharmaceutical Group Inc.
(2) For a report on µ-opioid receptor binding with relevant related compounds,
see: Schiller, P. W.; Nguyen, T. M.-D.; Chung, N. N.; Lemieux, C. J. Med.
Chem. 1989, 32, 698-703.
(3) Melchiorri, P.; Negri, L. Gen. Pharmacol. 1996, 27, 1099-1107.
(4) Erspamer, V. Int. J. DeV. Neurosci. 1992, 10, 3-30.
788
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development
10.1021/op020060k CCC: $22.00 © 2002 American Chemical Society
Published on Web 10/23/2002

the acid at low temperature (typically -10 to -25 °C). We
found that the mixed anhydrides formed in our process were
stable at -10 °C for 30 min up to several hours, an important
feature adding operability to the process. Following activa-
tion, the peptide coupling reaction is also fast. By addition
of the dissolved free-amino-block the bulk of the peptide
bond formation is done within a few minutes, and the
reaction is completed within 1 h. Yields of the coupling
product in the range of 95-98% were achieved. The
byproducts formed by this technique, carbon dioxide, isobu-
tanol, and NMM-hydrochloride, are easy to remove and
suitable for large-scale manufacture.
The mixed anhydride approach has been studied in detail
by several investigators,9-11 and our results follow most of
the findings described. However, DMF is suggested to be a
solvent where some racemisation can be expected, and the
stability of the mixed anhydride might not be the best. To
our benefit, DMF was found to be an excellent solvent when
used in the couplings described here, with long-lasting
stability of the mixed anhydrides and only trace amount of
racemisation in the peptide products. No other base than
NMM was evaluated for use in the frakefamide process.
From literature it is clear that NMM is a very suitable base
in combination with IBCF, especially with regards to low
racemisation.9 The mixed anhydride approach has earlier
been used for large-scale peptide synthesis,12 and examples
show that it is suitable for many peptide couplings;13
however, in some cases side reactions can be problematic.14
The nature of the amino acids/peptides to be coupled, and
careful optimisation of the conditions (temperature, solvent,
and stoichiometry) will determine the outcome. Clearly, the
mixed anhydride methodology suits the amino acid compo-
nents to be coupled in the frakefamide synthesis. That
together with thorough investigation of the conditions for
each coupling has provided good results.
Protecting-Group Strategy. Since activation with IBCF
and NMM occurs on the COOH function of the amino acid,
the amino group needs protection15,16 to prevent coupling
with itself. Any protecting group, and several can be
envisioned, therefore needs to be stable during activation and
the coupling reaction. Furthermore, dependable, selective,
and complete deprotection was a necessity to obtain clean
material for the next coupling and low overall impurity
formation. At the start of the development, the benzyloxy-
carbonyl group17 (CBz group) was chosen for amino protec-
tion. It can easily be removed by catalytic hydrogenolysis,
Figure 2. Mixed anhydride-promoted peptide coupling.
nontoxic). It should also be possible to remove residues of
the reagent conveniently post-coupling (e.g., by extraction
or filtration). A number of reagents fulfill these first important
requirements, but if a future production process is planned,
factors such as cost and availability, stability on storage,
environmental profile, and safety in use also have to be
considered.
After initial experiments with a few possible reagents,
further development work was focused around isobutyl
chloroformate (IBCF)-mediated mixed anhydride coupling
reactions.5-9 An example from the frakefamide × HCl (1b)
synthesis is shown in Figure 2. IBCF is a cheap and readily
available bulk reagent that together with the base N-
methylmorpholine (NMM) quickly (within minutes) activates
(10) Chen, F. M. F.; Lee, Y.; Steinauer, R.; Benoiton, N. L. Can. J. Chem. 1987,
65, 613-618.
(11) Chen, F. M. F.; Benoiton N. L. Can. J. Chem. 1987, 65, 619-625.
(12) For some general considerations of large-scale peptide synthesis in solution,
see: Gorup, B. Biochem. Soc. Trans. 1990, 18, 1299-1306.
(13) Brady, S. F.; Freidinger, R. M.; Paleveda, W. J.; Colton, C. D.; Homnick,
C. F.; Witter, W. L.; Curley, P.; Nutt, R. F.; Veber, D. F. J. Org. Chem.
1987, 52, 764-769.
(14) Prashad, M.; Prasad, K.; Repic, O.; Blacklock, T. J. Org. Process Res. DeV.
1999, 3, 409-415.
(15) Wu¨nsch, E. In Methoden der Organischen Chemie (Houben-Weyl); Mueller,
E. Eds., Thieme: Stuttgart, 1974; Vol. XV, part 2, pp 1-428.
(16) Geiger, R.; Ko¨nig, W. In The Peptides; Gross, E., Meienhofer, J., Eds.;
Academic Press: London, 1981; Vol. 3, pp 1-99.
(5) Vaughan, J. R. J. Am. Chem. Soc. 1951, 73, 3547.
(6) Wieland, T.; Bernhard H. Ann. 1951, 572, 190.
(7) Boissonnas, R. A. HelV. Chim. Acta 1951, 34, 874.
(8) Meienhofer, J. In The Peptides; Gross, E., Meienhofer, J., Eds.; Academic
Press: London, 1979; Vol. 1, pp 263-314.
(9) Anderson, G. W.; Zimmerman, J. E.; Callahan, F. M. J. Am. Chem. Soc.
1967, 89, 5012-5017.
(17) Bodansky, M.; Bodansky, A. In The Practice of Peptide Synthesis, 2nd ed.;
Springer-Verlag: Berlin, 1994; pp 11-14.
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
789

Figure 4. Process flow in a three-reactor unit.
Figure 3. Detrotection of CBz group by hydrogenolysis.
A was then transferred to reactor B where coupling occurred.
After completion of the coupling reaction, the newly formed
CBz-protected peptide was transferred to reactor C where
deprotection by hydrogenolysis was performed. The solid
Pd/C catalyst was removed by filtration with Filter X (a
pressure filter), and the solution with the amino free peptide
was transferred to reactor A. This reaction cycle was then
repeated for the introduction of the next amino acid. In
principle, the whole process is possible to run without any
isolation of intermediates. However, isolation is possible by
crystallisation after each reaction step by addition of an
antisolvent. To keep control of impurities and stoichiometry,
crystallisation and isolation of intermediates 9 and 1a by use
of Filter X was necessary.18 These intermediates, isolated
as solvent-moist crystals (only dry blown, not dried), were
taken out and weighed before recharging. By gaining more
experience of the process and better predictions of the yields,
it should not be necessary to empty the filter after these
isolations. The crystals can simply be redissolved in the filter
and transferred to the appropriate reactor for the next
reaction.
fulfilling all demands described above. The CBz group is
also a standard item in the art of peptide synthesis, making
sourcing of the desired CBz acids uncomplicated. In our
typical procedure (see Figure 3) the CBz group is removed
completely within 1-2 h by hydrogenolysis catalysed by
5% Pd/C at 3 atm. Similar to the coupling step, the
byproducts carbon dioxide and toluene are easy to remove
on large scale. Initially the Boc-group (tert-butyloxycarbonyl)
was also considered. It worked well as a protecting group
but its removal by TFA or HCl caused other problems. TFA
turned out to be difficult to remove from the final substance
without chromatography. If HCl was used instead, solvolysis
of the peptide amide started to occur, and the corresponding
ester was partly obtained. The CBz group had no side
reactions that caused similar problems. To our benefit, CBz
protection is also more economical due to lower reagent and
raw material costs.
Process Logistics. The overall process for a stepwise
synthesis of frakefamide hydrochloride (1b) (see Scheme 1)
consists of three couplings, three deprotections and a final
salt formation. Since the same procedures could be used
consecutively for each coupling and each deprotection step,
an attractive process design could be developed (see Figure
4). In a three-reactor unit, each reactor (A, B, and C
respectively) was utilised for a specific purpose. The amino
acid/peptide with free amino function to be coupled was
dissolved in reactor A, and the CBz-protected amino acid to
be added was charged to and dissolved in reactor B where
it was activated by IBCF and NMM. The content of reactor
Step-by-Step Analysis of the Synthesis (Scheme 1):
Step 1. Preparation of Dipeptide 5. In this first step, CBz-
protected L-p-fluorophenylalanine (2) is activated in DMF
by IBCF and NMM. Activation is performed at -10 °C,
whereupon the mixed anhydride 3 is formed. Under these
conditions the anhydride is stable for 3 h with negligible
racemisation. Activation can be performed in N-methyl-2-
(18) An additional filter was used for isolation of intermediates and final product
if efficient cleaning of Filter X from Pd/C catalyst could not be ensured.
790
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development

Scheme 1. Synthesis of frakefamide hydrochloride (1b)
pyrrolidone (NMP) and EtOAc with similar results, but DMF
was the best choice for the overall process. The peptide-
coupling reaction is performed at the same temperature by
addition of precooled phenylalanine amide (4) in DMF.
Coupling is fast, complete within a few minutes, and a
nonisolated yield of 5 of 98% is typically achieved. The DMF
slurry of 5 is used directly in the next step.
This step has been modified and enhanced considerably
during development. In the first pilot plant campaigns
performed, H-Phe-OMe (15) was used in place of H-Phe-
NH2 (4) (see Figure 5). The coupling product thus obtained,
dipeptide 16, was transformed by ammonolysis to the desired
dipeptide amide 5. Ammonolysis was, however, very time-
consuming, inconvenient, and volume-inefficient. Replace-
ment of H-Phe-OMe (15) by H-Phe-NH2 (4) gives an overall
performance benefit that compensates well for the higher
cost. In addition, intramolecular diketopiperazine formation
(see Figure 6) occurs much slower with a dipeptide amide
than with the corresponding ester. As a consequence, the
process becomes more robust, and the product contains less
of the piperazine impurity. Early on, dipeptide amide 5 was
isolated by crystallisation from DMF and water at 35-40
°C. The procedure is rather sensitive to the addition rate of
H2O, relative volumes of DMF-water and gives thin fibrous
crystals that are best isolated by use of a pressure filter. Better
control of the reaction conditions rendered this isolation
unnecessary.
Step 2. Hydrogenolysis To Obtain Dipeptide 6. The
second step, the first CBz group removal by hydrogenolysis,
is straightforward. The incoming slurry of CBz-protected
dipeptide amide 5 is warmed to 30 °C to improve solubility
and, as a result, the reaction rate. Catalytic Pd/C is added
and hydrogenolysis for 1-2 h at 3 atm removes the CBz-
group cleanly and completely. The catalyst is removed by
filtration and the obtained solution of 6 is used directly in
the next step. It can be noted that compound 6 is stable for
several days under the reaction conditions. For example, only
trace amounts of diketopiperazine is formed. The fact that
the incoming material for this step is not isolated has a
positive effect on the filtration of the catalyst. Most of the
N-methylmorpholine hydrochloride (NMM × HCl) formed
in the first step precipitates in the DMF phase and is filtered
off together with the catalyst. This makes the filter cake less
dense, and the filtration time is considerably shortened.
Different grades of Pd/C catalyst have been examined for
this reaction. For most pilot batches dry 5% Pd/C was used
and added to the reaction as a pre-made DMF slurry.
However, handling of dry Pd/C catalyst in regular manu-
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
791

(should be kept e-10 °C), is essential to avoid breakdown
and byproduct formation. However, if the temperature is too
low, a build-up of reagent will occur that may lead to a run-
away reaction. The coupling reaction is fast, as in step 1,
when dipeptide amide 6 is added to mixed anhydride 8. It is
finished within a few minutes with a nonisolated yield of
98%. The product 9 is crystallised by careful addition of
H2O, and the isolation is best performed with a pressure filter.
Centrifugation works less well due to the small particle size
of the crystals and high compressibility. Isolation is then
completed by acetone washes of the filter cake followed by
dry blowing with nitrogen. Dry blowing is efficient, leaving
only 2-3% of the solvent within the crystals, and no further
drying is needed prior to the next step. The isolated material
typically has a purity of 99% at this stage, and the overall
yield over the first three steps from 2 is about 93%.
Step 4. Hydrogenolysis To Obtain Tripeptide 10.
Hydrogenolysis of CBz-protected tripeptide amide 9 is
performed in the same way as for compound 5 in step 2. In
this case no NMM × HCl is present so that filtration of the
catalyst is much more time-consuming. The Pd/C-filter cake
becomes very dense and almost impenetrable. The reaction
time is about 1-2 h, and the resulting DMF solution of
tripeptide 10 is used directly in the next step. Both dry and
water-wet Pd/C catalysts have been utilised in this step with
very similar results.
Step 5. Preparation of Tetrapeptide Amide 13. This
step, the final peptide coupling where CBz-tyrosine 11 is
added to the tripeptide 10, is the most delicate in the
synthesis. A small excess of CBz-tyrosine (11) is used and
activated by IBCF/NMM in DMF in the same manner as
described in steps 1 and 3. DMF and NMP work fine for
the activation, but stability of the mixed anhydride 12 was
poor in EtOAc. The outcome of the reaction is sensitive to
the relative amounts of IBCF and NMM charged, and
equimolar amounts of the reagents and CBz-tyrosine should
be used. Too little reagents charged means, of course,
incomplete formation of the mixed anhydride 12, but only
slight overcharge will promote isobutyl carbonate acylation
of the tyrosine phenolic hydroxyl group (see Figure 7). If
formation of the acylated byproduct 17 occurs, peptide
coupling will still be possible, and a tyrosine acylated
tetrapeptide 18 will be formed as a coupling byproduct. This
byproduct can be converted to the desired product 13 by
ammonolysis. In essence, if a significant amount of byprod-
uct 18 is formed, ammonia treatment is included in the
workup procedure. Crystallisations later in the synthesis
remove this byproduct poorly so that the development of
the optional ammonia treatment made the overall process
more robust.
Figure 5. Previous synthesis of dipeptide 5.
Figure 6. Diketopiperazine formed at dipeptide stage.
facture is not desirable from a safety point of view. Therefore
water-wet catalyst (10% Pd/C, 50% H2O content) was
examined both in the laboratory and the pilot plant. The
results were equivalent with those obtained with a dry
catalyst with perhaps a small increase of the reaction rate.
How much water that can be tolerated in the overall process
remains to be investigated. Initial studies on the presence of
water in the peptide couplings as well as literature prece-
dence10,19 are promising.
Step 3. Preparation of Tripeptide Amide 9. Activation
of CBz-D-Ala-OH (7) with IBCF and NMM is best per-
formed in acetone or acetonitrile. Both offer good stability
(6 h at -10 °C) of the mixed anhydride 8 with very low
rate of racemisation. Acetone was used in the pilot plant as
the cheaper and more environmentally friendly choice.
Control of temperatures, especially during the IBCF charge
In early development, compound 13 was crystallised by
addition of H2O and isolated. This crystallisation was very
sensitive and gave crystals of poor quality that were very
difficult to filter due to partial gel formation. Experience gave
better control of the reaction, and isolation at this stage could
be omitted. However, the obtained solution of 13 was not
suitable for the next step since the presence of NMM × HCl
would greatly disturb the crystallisation of frakefamide (1a)
(19) Chen, F. M. F.; Steinauer, R.; Benoiton, N. L. J. Org. Chem. 1983, 48,
2939-2941.
792
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development

precipitation of product can be avoided. This will also
facilitate efficient DMF-wash of the filter cake that is
necessary to achieve a good yield. After Pd/C filtration,
frakefamide (1a) is crystallised by addition of EtOAc and
isolated by filtration. The crystals are blow-dried and are
used EtOAc-wet in the next step. A solvent content up to
25% (w/w) can be tolerated. Removal of NMM × HCl in
the previous step, as well as ammonia treatment to remove
any tyrosine-O-acylated product, together with the hot Pd/C
filtration have made this step more robust and better yielding.
The isolated yield of frakefamide 1a is about 80% over two
steps starting from tripeptide 9, with majority of the loss
occurring in the crystallisation of product 1a.
Step 7. Formation of Frakefamide × HCl (1b). The
free base 1a is charged EtOAc-wet and is dissolved in a
H2O-acetone mixture. HCl is charged (1 equiv), and the
solution is clear filtered. When more acetone and seed
crystals are added, the crystallisation starts, and addition of
MIBK then completes crystallisation. The temperature is kept
at 20 °C throughout the procedure, and the crystals are
preferably isolated with a pressure filter. The yield is good,
about 95%, and a purity of 99.5% of the final product 1b
can be expected, a purity improvement of 3% over the step.
If this procedure is followed, frakefamide × HCl (1b) is
isolated in a stable crystal form. However, if the crystals
are left MIBK-wet for more than a week, other forms and
amorphous material will start to appear. The dried product
is, on the other hand, stable, only weakly hygroscopic, and
has a shelf life exceeding 2 years.
Investigation of a Convergent Alternative. An alterna-
tive route was developed where two dipeptides were coupled
to yield tetrapeptide 13 as an alternative to the step-by-step
synthesis described above (Scheme 2). The synthesis was
performed up to 2-L scale. The first fragment, dipeptide 6,
was synthesised as described earlier. The next building block
CBz-Tyr-D-Ala-OH (20) was also synthesised using the
IBCF/NMM coupling technique. CBz-Tyr-OH (11) and H-D-
Ala-OMe × HCl (19) were starting monomers, and the
coupling could be performed entirely in acetone. NaOH was
subsequently used to remove the methyl ester, and the
dipeptide fragment 20 was ready for coupling. Dipeptide
fragment 20 could also be prepared from free D-alanine to
avoid the acid deprotection step. A poly(ethylene glycol)-
EtOAc mixture was found to be efficient, and K2CO3 was
added to ensure desired deprotonation of the D-alanine.
Formation of dipeptide 20 by this route was promising, but
isolation turned out to be more problematic.
The two peptide fragments were coupled using the same
IBCF-promoted mixed anhydride approach. A slight variation
of the procedure was found to be efficient.20 Both compo-
nents (6 and 20) were charged to a reactor, and IBCF was
added at -20 °C. Peptide-bond formation was then promoted
by controlled addition of NMM. Also in this case, the whole
procedure could be performed in acetone, and tetrapeptide
13 was isolated in 75-80% yield.
Figure 7. Formation and removal of byproduct 18.
to be produced in the next step. To circumvent this problem,
one-third of the DMF is replaced with EtOAc after which
NaHCO3 is added. This converts NMM × HCl into neutral
NMM and NaCl. The NaCl precipitates in the EtOAc-DMF
mixture and can easily be filtered off. The NMM is distilled
off at a later stage.
Step 6. Hydrogenolysis To Obtain Frakefamide (1a).
CBz-protected tetrapeptide 13, incoming as a DMF-ETOAC
solution, is hydrogenolysed as described for steps 2 and 4
with formation of the unmasked product 1a. The reaction
time is about 1-2 h. Filtration of the Pd/C-catalyst is best
performed with a heated filter (30-40 °C) so that unwanted
(20) A similar activation procedure has been described by: Shieh, W.-C.; Carlson,
J. A.; Shore M. E.; Tetrahedron Lett. 1999, 40, 7167-7170.
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
793

Scheme 2. Synthesis of tetrapetide 13 using dipeptide
building blocks
decrease the cost due to lower raw material prices. Of the
raw materials, CBz-Phe(F)-OH (2) is the most costly
followed by H-Phe-NH2 × HCl (4). These two components
are about 5-10 times more expensive than CBz-D-Ala-OH
(7) and CBz-Tyr-OH (11). Reagents and solvents will only
add negligible sums to the raw material cost. In an overall
estimate, raw materials contribute 60-70% of the cost of
goods, and the remaining part is used for operation of the
process.
Conclusions
Starting from amino acid monomers, the overall yield in
the tetrapeptide synthesis over seven steps of frakefamide
× HCl (1b) is just over 70%. That together with the high
purity (99.5%) of the final substance is in itself a good
measure of the standard of this process. The chemistry that
is used, IBCF/NMM-promoted couplings and CBz-group
deprotection by hydrogenolysis is straightforward and simple.
It is also robust, high-yielding, and provides very little
formation of undesired byproducts. For example, there is very
little racemisation (only traces detected) of the amino acid
components occurring during the peptide-coupling steps.
Crystallisations of intermediates and final substance are
somewhat sensitive but provide good purification and yields.
Experimental Section
Structure-Contributing Components. All amino acid
components were obtained from commercial sources, H-Phe-
NH2 × HCl (4) from Senn Chemicals AG, CBz-Phe(F)-OH
(2) from Synthetech Inc., CBz-D-Ala-OH (7) and CBz-Tyr-
OH (11) from Rexim Degussa.
Reagents. IBCF (SNPE Chemie), NMM (BASF), and
Pd/C (5% Pd/C type 5R87L, from Johnson Matthey) were
purchased from the listed commercial sources in reagent
grade. Other chemicals were standard bulk chemicals.
Solvents. DMF was purchased in drums (p.a. quality from
Merck). Acetone, water, MIBK, and EtOAc were of bulk
quality taken from the local plant solvent supply system.
Equipment. Reactor A and reactor B were both 70-L
glass-lined reactors, each equipped with a three-bladed retreat
curve impeller. Reactor C was a 70-L stainless steel reactor
equipped with a 12-bladed turbine. Filter X was a pressure
filter-nutsche equipped with a K200-filterplate.18
Analytical Methods. A generic LC-purity method was
developed that was used for all steps. The method was used
both for in-process control and for determination of the purity
of isolated substance. The degree of purity was determined
by comparison with reference standards of known purity.
Correct identity of intermediates and isolated material were
controlled by comparison with reference standards, which
were characterised by NMR and MS.
Although the convergent route was not developed to the
same extent as the linear synthesis, its potential could be
estimated and the two alternatives compared. Yield, quality,
and cost of final substance were comparable from both
routes, but batch size could be larger using the convergent
approach. On the other hand, process logistics suggest more
efficient use, and shorter batch cycle times, for the linear
synthesis. The potential for outsourcing of the dipeptide
blocks 6 and 20 and less use of DMF were other benefits
with the convergent approach. In light of the relatively small
differences between the routes and more expertise with the
linear one, further development of a convergent approach
was not carried out.
Resources and Costs. This process was developed during
a period of 4 years during which seven pilot-plant campaigns
were carried out. A total of about 20 man-years of effective
development time was used, including all aspects of process
development. Estimation of the cost of frakefamide × HCl
(1b) by the described process is $2500/kg for production of
less than 1000 kg/year. Higher annual production would
The Generic LC-Purity Method. Column: Kromasil C8,
3.5 µm, 100 mm × 4.6 mm or equivalent; flow rate: 1.0
mL/min; injection volume: 10 µL; wavelength: 210 and 254
nm; run time: 18.5 min; mobile phase A: 0.1% (v/v) TFA
in H2O; mobile phase B: 0.09% (v/v) TFA in 80/20 (v/v)
CH3CN/H2O.
794
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development

Gradient
time
°C. After the agitation was lowered and the hydrogen
pressure was released, the mixture was sampled. LC-purity
analysis showed that 99.7% of 5 (based on total area %
calculations) had been consumed and that content of 6 was
96.3 area %. (If conversion of 5 had been lower than 98%,
new H2 would have been charged to 3 atm pressure and the
reaction continued.) The content of reactor C was filtered
by Filter X and the filtrate collected in reactor A. DMF (8
L) was used to rinse reactor C and wash Filter X; the wash
liquid was collected in reactor A. The solution, containing
6, was cooled to -10 °C and used directly in the next step.
Simple isolation of smaller amounts is possible by
evaporation and drying of the residue. These operations gave
6 as a solid powder, not further purified but solvent-free,
which had NMR data: 1H NMR (DMSO-d6) δ 1.65 (s
(broad), 2H) 2.47-2.53 (m, 1H), 2.74-3.00 (m, 3H), 3.93
(m, 1H) 4.47 (m, 1H), 7.02-7.43 (m, 11H), 7.99 (d, J )
6.9 Hz, 1H)
CBz-D-Ala-Phe(F)-Phe-NH2 (9). CBz-D-Ala-OH (7) (1.42
kg, 6.35 mol) and acetone (5.8 kg, 7.3 L) were charged to
reactor B. The mixture was stirred at 25 °C, and after 15
min a thin opaque solution was obtained, and the temperature
could be lowered to -10 °C. Isobutyl chlorofomate (IBCF)
(0.87 kg, 826 mL, 6.35 mol) was then charged to reactor B
over 2 min. This charge was directly followed by a charge
of NMM (0.74 kg, 804 mL, 7.32 mol) over 20 min, while
the reaction temperature was kept at -5 to -15 °C.
(CAUTION: Addition of NMM causes an exothermic
reaction, and temperature needs to be carefully monitored.)
Without delay, the entire DMF solution of 6 in reactor A
(obtained in previous step) was transferred to reactor B over
20 min; the reaction temperature was kept at -5 to -10 °C.
(CAUTION: The peptide-coupling reaction is exothermic,
and temperature needs to be carefully monitored.). Reactor
A was washed with DMF (3 L) which then was transferred
to reactor B. After 30 min the reaction mixture was sampled;
the temperature of the reaction mixture was adjusted to 20
°C. LC-purity analysis of the sample showed that all 6 was
consumed and that 9 was present in 92.8 area %. The reaction
mixture was left overnight and then distilled at reduced
pressure (100 mbar, temperature of jacket 50 °C). About half
the volume was removed (∼20 L) by the distillation. Acetone
(8 L) was charged to the remaining thin slurry and the
temperature was adjusted to 40 °C. Over 1 h, water (41 L)
was then charged to reactor B in portions; the temperature
was kept between +30 to +40 °C. After completed H2O
charge the crystal slurry was gently stirred overnight and
the temperature adjusted to 20 °C. The crystals were isolated
by filtration with Filter X and then washed at 20 °C with a
water-acetone mixture (1:1, 5 L) followed by two identical
acetone washes (5 L each). After blow-drying with N2 (100
min) the filter was emptied, and 3.34 kg of 9 was obtained.
The solvent content was 2.5%, and the LC purity of 9 in the
isolated material was 99.0%. A calculation of the yield would
then give 3.03 kg of pure 9 that correlates to 5.7 mol and
93% yield over three steps starting from 2. The crystals of
9 had NMR data: 1H NMR (DMSO-d6) δ 0.97 (d, J ) 6.8
Hz, 3H), 2.66-3.08 (m, 4H), 4.02 (m, 1H), 4.45 (m, 2H),
%A
%B
0
2.5
12.5
16
16.5
18.5
75
75
10
10
75
75
25
25
90
90
25
25
1H NMR spectra were recorded on a Bruker Avance 400
(400 MHz) or a Varian Gemini 300 (300 MHz) at 25 °C.
The samples were dissolved in DMSO-d6, and the DMSO
peak at δ 2.50 ppm was used as internal standard.
CBz-Phe(F)-Phe-NH2 (5). H-Phe-NH2 × HCl (4) (1.27
kg, 6.33 mol) and DMF (6.3 kg, 6.7 L) were charged to
reactor A at 20 °C. After complete dissolution, the temper-
ature was adjusted to -10 °C. NMM (0.66 kg, 719 mL, 6.53
mol) was then charged to the solution to neutralise the
hydrochloride salt. CBz-Phe(F)-OH (2) (1.94 kg, 6.10 mol)
and DMF (6.3 kg, 6.7 L) were charged to reactor B at 20
°C. After almost complete dissolution, the temperature in
reactor B was adjusted to -10 °C, and isobutyl chlorofomate
(IBCF) (0.95 kg, 902 mL, 6.95 mol) was charged over 2
min. This charge was directly followed by a charge of NMM
(0.71 kg, 774 mL, 7.02 mol) over 15 min, and reaction
temperature was kept at -5 to -15 °C. (CAUTION:
Addition of NMM causes an exothermic reaction, and
temperature needs to be carefully monitored.) Without delay,
the entire contents of reactor A were transferred to reactor
B over 10 min; the reaction temperature was kept at -5 to
-10 °C. (CAUTION: The peptide-coupling reaction is
exothermic, and temperature needs to be carefully moni-
tored.) Reactor A was washed with DMF (3 L) which then
was transferred to reactor B. The reaction mixture was stirred
at -10 °C and sampled after 20 min. Analysis showed LC-
purities of 97.2 and 0.9 area % of 5 and 2, respectively. (If
the level of 2 had been over 4 area %, reaction time would
have been prolonged or additional IBCF charged or both).
The thin DMF slurry of 5 in reactor B was used directly in
the next step.
Crystallisation of dipeptide 5 is possible by slow addition
of water at 20-30 °C. The formed crystals can then be
isolated by filtration and dried. Dipeptide 5 isolated by this
procedure gave NMR data: 1H NMR (DMSO-d6) δ 2.62-
3.07 (m, 4H), 4.22 (m, 1H), 4.48 (m, 1H), 4.94 (m, 2H),
7.0-7.54 (m, 17H), 8.05 (d, J ) 8.2 Hz, 1H).
H-Phe(F)-Phe-NH2 (6). The DMF slurry of 5, obtained
in the previous step, was transferred from reactor B to reactor
C. Reactor B was washed with DMF (4 L) which then was
transferred to reactor C, and the inner temperature was
adjusted to 30 °C. Pd/C (143 g, 5% Pd content, dry) was
mixed with DMF (420 mL) in a steel vessel, and the slurry
was charged to reactor C via the manhole. H2 (g) was charged
to reactor C until pressure reached 3 atm (0.3 Mpa), and the
agitation was adjusted to strong. Within 45 min, the hydrogen
pressure was released and new hydrogen gas charged twice
to the same pressure. The mixture was kept for 2 h with
strong agitation while temperature was allowed to fall to 25
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
795

5.01 (q, 2H), 7.00-7.46 (m, 17H), 8.04 (d, J ) 8.2 Hz, 1H),
8.13 (d, J ) 8.3 Hz, 1H)
mL, 0.04 mol) and NMM (29 mL, 0.05 mol) were made.
After 10 min of continued stirring the temperature was
adjusted to 20 °C, and the mixture was left for an additional
2 h. (If the sample had shown formation of byproduct 18
(>0.2 area % in LC-purity test) charge of 2.33 L NH4OH
(25% aqueous solution) would have been necessary. Heating
of the obtained mixture at 60 °C for 4 h under strong agitation
removes byproduct 18, and the process can proceed with
the next operation.) After 18 L of DMF was removed by
distillation performed at jacket temperature of 50 °C and
reduced pressure, EtOAc (20 L) was charged. NaHCO3 (1.14
kg, 17.1 mol) was charged via the manhole, and the mixture
was left with powerful stirring for 4 h at 30 °C. The
temperature was then adjusted to 5 °C, and the mixture was
filtered (precipitated NaCl isolated on filter) with filter X to
reactor A. A DMF-EtOAC mixture (1:1, 4 L) was used to
rinse reactor B and wash Filter X; the wash liquid was
collected in reactor A. The DMF solution of 13 was used
directly in the next step. LC-purity analysis of the solution
showed a content of 13 of 91.2 area % (by calculation this
correlates to 3.6 kg and 5.6 mol).
Crystallisation of tetrapeptide 13 is possible by slow
addition of CH3CN (0.7 L/ L DMF) at 45 °C followed by
seeding with good quality crystals. The temperature is then
slowly lowered to 20 °C before a slow addition of water
(0.7 L/ L DMF). The crystals can then be isolated by filtration
and dried. Tetrapeptide 13 isolated by this procedure has
NMR data: 1H NMR (DMSO-d6) δ 0.90 (d, J ) 6.9 Hz,
3H), 2.51-3.03 (m, 6H), 4.21 (m, 2H), 4.45 (m, 2H), 4.93
(q, 2H), 6.64 (d, J ) 8.2 Hz, 2H), 7.00-7.45 (m, 19H), 8.04
(d, J ) 7.3 Hz, 1H), 8.09 (d, J ) 8.2 Hz, 1H), 8.17 (d, J )
8.4 Hz, 1H), 9.2 (s, 1H).
Frakefamide (H-Tyr-D-Ala-Phe(F)-Phe-NH2) (1a). The
DMF solution of 13 (3.91 kg, 5.6 mol) was transferred from
reactor A to reactor C; DMF (4 L) was used to rinse reactor
A. The inner temperature was adjusted to 30 °C. Pd/C (234
g, 5% Pd content, dry) was mixed with DMF (940 mL) in a
steel vessel, and the obtained slurry was charged to reactor
C via the manhole. Hydrogen (g) was charged to reactor C
until pressure reached 3 atm (0.3 Mpa), and agitation was
adjusted to strong. Within 45 min, the hydrogen pressure
was released, and new hydrogen gas charged twice to the
same pressure. The mixture was kept for 1.5 h with strong
agitation. After adjustment to lower agitation and release of
the hydrogen pressure, the mixture was sampled. LC-purity
analysis showed that 99.8% of 13 (based on total area %
calculations) had been consumed. (If conversion of 13 had
been lower than 97%, new hydrogen would have been
charged to 3 atm pressure and reaction continued at 40 °C
for 30 min.) The content of reactor C was filtered at 35 °C
by Filter X, and the filtrate was collected in reactor A. DMF
(4 L) was used twice to rinse reactor C and wash Filter X;
the wash liquids were collected in reactor A. Both washes
were performed at 35 °C. Twenty-three liters of DMF was
removed by distillation performed at jacket temperature of
50 °C and reduced pressure. EtOAc (7.8 + 15 + 28 L) was
charged in portions at 30-40 °C over 2 h. Crystals forms
spontaneously after the first portion. Once all EtOAc was
H-D-Ala-Phe(F)-Phe-NH2 (10). Tripeptide 9 (3.3 kg of
material isolated in the previous step, correlates to 3.0 kg,
5.7 mol dry and pure material) and DMF (23 L, 17 kg) was
charged to reactor B at 20 °C. The jacket temperature was
then set to 45 °C and vacuum was applied. Under these
conditions about 8 L of solvent was removed to ensure that
all acetone was removed. After distillation the vacuum was
released, and the thin slurry was transferred to reactor C;
DMF (4 L) was used to rinse reactor B. The inner
temperature was adjusted to 30 °C. Pd/C (153 g, 5% Pd
content, dry) was mixed with DMF (610 mL) in a steel
vessel, and the obtained slurry was charged to reactor C via
the manhole. H2 (g) was charged to reactor C until pressure
reached 3 atm (0.3 Mpa), and the agitation was adjusted to
strong. Within 45 min, the hydrogen pressure was released
and new hydrogen gas charged twice to the same pressure.
The mixture was kept for 1.5 h with strong agitation. After
adjustment to lower agitation and releasing of the hydrogen
pressure, the mixture was sampled. LC-purity analysis
showed that 99.9% of 9 (based on total area % calculations)
had been consumed and that the content of 10 was 95.3 area
%. (If conversion of 9 had been lower than 98%, new H2
would have been charged to 3 atm pressure and the reaction
continued.) The content of reactor C was filtered by Filter
X and the filtrate collected in reactor A. DMF (4 L) was
used to rinse reactor C and wash Filter X, the wash liquid
was collected in reactor A. The solution, containing 10, was
cooled to -10 °C and used directly in the next step.
Simple isolation of smaller amounts is possible by
evaporation and drying of the residue. These operations gave
10 as a solid powder, not further purified but solvent-free,
1
which had NMR data: H NMR (DMSO-d6) δ 0.96 (d, J )
6.8 Hz, 3H), 1.73 (s (broad), 2H) 2.73-3.05 (m, 4H), 3.18
(q, J ) 6.9 Hz, 1H), 4.46 (m, 2H), 6.99-7.40 (m, 11H),
8.04 (s (broad), 1H), 8.11 (d, J ) 8.3 Hz, 1H).
CBz-Tyr-D-Ala-Phe(F)-Phe-NH2 (13). CBz-Tyr-OH (11)
(1.87 kg, 5.92 mol) and DMF (7.7 kg, 10 L) were charged
to reactor B. The mixture was stirred at 25 °C and after 15
min a clear gray solution was obtained, and the temperature
could be lowered to -10 °C. Isobutyl chlorofomate (IBCF)
(0.808 kg, 770 mL, 5.92 mol) was then charged to reactor
B over 2 min. This charge was directly followed by a charge
of NMM (0.604 kg, 659 mL, 5.97 mol) over 15 min; reaction
temperature was kept at -5 to -15 °C. (CAUTION:
Addition of NMM causes an exothermic reaction, and
temperature needs to be carefully monitored.) Without delay,
the entire DMF solution of 10 in reactor A (obtained in the
previous step) was transferred to reactor B over 10 min; the
reaction temperature was kept at -5 to -10 °C. (CAU-
TION: The peptide-coupling reaction is exothermic, and
temperature needs to be carefully monitored.) Reactor A was
washed with DMF (3 L) which then was transferred to
reactor B. After 30 min the reaction mixture was sampled.
LC-purity analysis of the sample showed that 4.2 area % of
10 was unreacted. Since internal criteria for complete reaction
was <1 area % of remaining 10, extra charges of IBCF (31
796
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development

added, the slurry was cooled to 5 °C within 30 min and then
left with stirring for 2 h. The crystals were isolated by
filtration with Filter X, and the crystals were washed twice
with EtOAc (5+5 L) and blow-dried with N2. The filter was
emptied, and the crystals were dried in a drying cabinet at
50 °C and 20 mbar for 100 h; 3.20 kg frakefamide (1a) was
finally isolated. The solvent content of 1a was 22% (mainly
EtOAc), and LC-purity was 97.6 area %. This correlates to
a dry and pure weight of 2.48 kg (4.41 mmol) and a yield
and methyl isobutyl ketone (MIBK) was charged in four
portions (4 × 14 L) to reactor B. Each portion was charged
with a rate of about 0.4 L/min with a 10-15 min break and
slow stirring between each portion, the temperature was kept
at 20 °C. After the MIBK charge was completed, the crystal
slurry was stirred for 1 h and then filtered using Filter X.
The crystals were washed twice with MIBK (8 + 8 L) and
then blow-dried with N2. The moist crystals were emptied
from the filter and dried in a drying cabinet for 150 h at 50
°C and <15 mbar. Finally, dry crystals (solvent + H2O
content was 0.3%) of 1b (2.52 kg, 4.18 mmol) could be
collected. LC purity of the material was 99.5%, and the yield
from 1a, 96%. The obtained crystals of 1a had NMR data:
1H NMR (DMSO-d6) δ 0.74 (d, J ) 6.9 Hz, 3H), 2.60-
3.06 (m, 6H), 3.95 (m, 1H), 4.27 (m, 1H), 4.41 (m, 2H),
6.68 (d, J ) 8.2 Hz, 2H), 6.95-7.32 (m, 12H), 7.51 (s, 1H)
8.23 (s (broad), 3H), 8.36 (m, 2H), 8.48 (d, J ) 8.2 Hz,
1H), 9.39 (s, 1H).
1
of 80% from 9. H NMR data correlates to those obtained
for compound 1b, except that the signal from aminoprotons
will vary in δ value, depending on pH.
Frakefamide Hydrochloride (H-Tyr-D-Ala-Phe(F)-Phe-
NH2 × HCl) (1b). Aqueous HCl (37% w/w, 510 g, 5.17
mmol), water (249 g, 249 mL), and acetone (7.9 kg, 10 L)
were added to a charge vessel and mixed at 20 °C.
Frakefamide (H-Tyr-D-Ala-(F)Phe-Phe-NH2) (1a) (3.166 kg
EtOAc moist, obtained in the previous step correlates to 4.38
mmol) was charged to reactor A followed by a rapid charge
of the acetone-hydrochloride mix from the charge vessel.
Stirring for 15 min at 20 °C gave an almost clear solution
that was clear filtered, using a polish filter, to reactor B.
Reactor A and the filter was washed with an acetone-water
mixture (1.37 L acetone + 0.071 L water, 95:5 v/v) and the
wash liquid added to the content in reactor B. Stirring was
adjusted to slow and acetone (10.5 L) was charged while
the temperature was kept at 20 °C. Good quality frakefamide
Acknowledgment
The following AstraZeneca collaborators have provided
their expertise and made invaluable contributions to the
process described herein: Thony Harnesk, Peter Koolmeister,
Krister Mellstro¨m, Dr. Eva Jakobsson, Mårten Ellburg,
Richard Kemp, David Fletcher, Petra Gro¨nvall, Ilse Hen-
riksson, Dr. Per-Ivar Gunnarsson, Petra Zunko, Dr. Karol
Horvath, Jeroen Konigen, Homa Amini, Dr. Anne Ertan, Dr.
Mats Andersson, and Dr. Hans-Ju¨rgen Federsel.
hydrochloride crystals (H-Tyr-D-Ala-(F)Phe-Phe-NH2
×
HCl) (1b) (30 g, 0.05 mol) and acetone (535 mL) were mixed
to a seed slurry in a flask. The seed slurry was carefully
charged to reactor B via the manhole which initiated
crystallisation. After 20 min, stirring was adjusted to strong,
Received for review July 1, 2002.
OP020060K
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
797