Development and scale-up of an optimized route to the peptide boronic acid, CEP-18770

Article abstract of DOI:10.1021/op400010u

CEP-18770 is an unstable peptide boronic acid and an amorphous solid, making it a challenging synthetic target. Process R&D led to a new process that avoided chromatography through crystalline intermediates, increased atom and volume efficiency, provided

Full text of DOI:10.1021/op400010u

Article
pubs.acs.org/OPRD
Development and Scale-Up of an Optimized Route to the Peptide
Boronic Acid, CEP-18770
Renee C. Roemmele* and Michael A. Christie
Chemical Process Research and Development, Teva Branded Pharmaceutical Products R&D Inc., 383 Phoenixville Pike, Malvern,
Pennsylvania 19355, United States
ABSTRACT: CEP-18770 is an unstable peptide boronic acid and an amorphous solid, making it a challenging synthetic target.
Process R&D led to a new process that avoided chromatography through crystalline intermediates, increased atom and volume
efficiency, provided a chromophore, and gave higher yields and purity. A stable, crystalline diethanolamine adduct was discovered
that has the potential to be used as a prodrug.
intermediate 7 that can be purified by recrystallization, if
needed, and eliminated a deprotection step from the sequence
by avoiding the use of a protected threonine derivative.
In step 2, TBTU was used as an activating agent along with
NMM during preparation of preclinical supplies. To a cooled
solution of 7, NMM, and TBTU was co-fed a second equivalent
of NMM and 1 as the HCl salt. Although the yields were high, a
chromatographic purification was required to remove the 1−3%
of the undesired diastereomer that resulted from epimerization
during coupling. Later, during preparation of the phase 1
supplies, improved peptide coupling conditions reduced
epimerization to <1%. In the improved coupling, 1 and 7
were first dissolved in DMF along with HATU as the coupling
reagent. The mixture was then cooled below −25 °C, followed
by slow addition of DIPEA as base. A carbon treatment of an
ethyl acetate solution of 5 was key to increasing the chemical
purity, allowing for isolation of the product as a clear, pale-
yellow foam with >96% purity without chromatography.
In the final step, a transesterification with isobutyl boronic
acid to remove the pinane diol ester was carried out under acid
catalysis using a two-phase MeOH/heptanes mixture.2 Crude
delanzomib remained in the methanol phase, and upon
isolation had to be purified by chromatography. Finally, the
resultant amorphous solid was triturated with either tert-butyl
methyl ether (TBME) or pentane to obtain a flowable solid. A
low-boiling solvent was necessary, as delanzomib degrades
quickly above 30 °C or when exposed to light or air, making it a
challenge to isolate, purify, and dry.
INTRODUCTION
■
Delanzomib (CEP-18770, 6) is a peptide boronic acid whose
main indication is multiple myeloma and which acts as a
reversible inhibitor of the proteasome.1 It is a challenging
synthetic target, as it is an amorphous solid that is both
unstable, requiring storage at −20 °C, and very potent with an
occupational exposure limit of 0.3 μg/m3. The original
synthesis, described in 2005 and outlined in Scheme 1,
required four steps and gave an overall yield of 25%. It
suffered from several drawbacks, including the lack of a
chromophore for monitoring the first two reactions, and there
were no crystalline intermediates.2 Purification of intermediates
and the final product was difficult and required column
chromatography.
In addition, the key pinane diol boronic ester starting
material, 1,3 was difficult to obtain in high chiral purity, and
there was no effective means of improving the optical purity of
downstream intermediates or the final product. Using the
discovery synthesis, delanzomib could be obtained in 96−98%
purity with ∼98% de.
The new, optimized process addresses these shortcomings
and employs a novel crystalline diethanolamine adduct to
upgrade both the chemical and optical purity of the final
product. This adduct, which can be obtained with 99.9% purity
and >99.8% de, also has the potential to be used as a prodrug.
Process Development. The peptide boronic acid can be
built starting with the boronic acid terminus, as in the original
synthesis, or starting from the amino terminus. The latter
approach not only introduces a chromophore for ready in-
process monitoring and analysis of all steps, it also allows for
later introduction of the “warhead” boronic acid, limiting the
number of steps requiring potent compound handling and
introducing the expensive boronic acid moiety toward the end
of the synthesis. A disadvantage to this approach is the increase
in potential for epimerization during amide bond formation.4
Preclinical and initial phase 1 supplies were synthesized using
the process outlined in Scheme 2. In step 1, the acid chloride of
4 is prepared using thionyl chloride and then used directly in a
Schotten-Bauman reaction with L-threonine to form 7. The
product is crystallized from the aqueous phase following
addition of methanol and acidification. In addition to
introducing a chromophore, this route provided a crystalline
The new process was an improvement over the original
synthesis. However, the following issues remained: the need for
column purification of the final product; the reliance on the
optical purity of 1 to set the final % de; the existence of only
one crystalline intermediate, early in the sequence; and an
unstable, amorphous API that could only be obtained in
moderate (94−97%) purity.
Boronic acids are known to form crystalline trimeric
anhydrides,5 but all efforts to generate a trimer of delanzomib
were unsuccessful. Extensive screening efforts did not uncover a
Received: January 14, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op400010u | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Original synthetic process
Scheme 2. Preclinical and initial phase 1 process
crystalline form of the base nor an isolable salt. However, a
literature survey indicated that amino bis-hydroxyalkyl and
amino bis-carboxyalkyl compounds can form crystalline adducts
of boronic acids.6 Table 1 summarizes our screening efforts.
Diethanolamine (DEA, entry 1) was best, providing a
crystalline adduct that also upgraded both chemical and optical
purity. Diisopropanolamine (entry 4) also provided a crystalline
material, but yields were lower, and a mixture of diastereomers
was obtained.
When a solution of crude delanzomib in ethyl acetate was
treated with DEA, a crystalline adduct formed almost
immediately. Alternatively, 8 could be prepared directly from
intermediate 5 (Scheme 3) by acid-catalyzed transesterification,
albeit in lower yield and with significantly longer cycle times.
Characterization of 8 revealed it to be the desired diastereomer
Table 1. Derivative screening
Scheme 3. Formation of DEA adduct 8
entry
R
X, Y
n, m
result
1
2
3
4
5
6
7
8
H
H, H
1, 1
1, 1
2, 2
1, 1
2, 2
1, 2
1, 1
1, 1
crystalline isomer upgrade
amorphous solid
H
CO
H
H, H
no solids isolated
H
H, CH3
CO
H, H
crystalline isomer upgrade
amorphous solid
H
H
no solids isolated
CH3
CH3
H, H
amorphous solid
CO
no solids isolated
B
dx.doi.org/10.1021/op400010u | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
in >99.9% de. A single-crystal X-ray structure (Figure 1)
confirmed it as a monohydrate. It can be recrystallized from
prodrug for oral administration. Unfortunately, variable rates of
uptake dependent on the individual were observed along with
increased liver toxicity. The oral dosing program was
abandoned, but not the idea of using 8 as a prodrug.
Compound 8 proved to be a direct substitute for delanzomib
in the formulation process. In the first step of the IV
formulation process, delanzomib is dissolved in water along
with several excipients. Predictably, the delanzomib degrades
during this process. It was found that upon dissolution in the
lyophilization medium, 8 hydrolyzes to delanzomib, but the fact
that delanzomib forms in situ, coupled with both the limited
handling time and higher initial purity, meant that the drug
product was of much higher purity. Moving into phase 3, the
drug substance was changed to 8, but the active ingredient in
the drug product remained delanzomib. Using the formulation
process as the final step in the synthesis of the API is a unique
approach. In addition, there is no longer the need to generate
many small batches of delanzomib, as 8 can be made in large
batches and stored for several years.
Phase 3 Process. Leading up to the phase 3 campaign,
several design of experiments studies (DoEs) were run to
determine and optimize the operating ranges of each variable in
each step of the process. Design Expert software was used to
evaluate the data and create a visual representation of the
boundaries of each reaction. Scheme 4 summarizes the final
process used for the preparation of phase 3 supplies.
The DoE studies led to some interesting results. We found in
the acid chloride formation that the amount of thionyl chloride
could be reduced from 5 to 2 equiv by running the reaction at
higher concentrations. For the Schotten−Bauman reaction, the
only significant factors were that the equivalents of L-threonine
initially charged had to exceed the equivalents of acid chloride,
and to avoid formation of a difficult-to-remove impurity, a
minimum of 1.3 equiv of sodium carbonate was required.
Finally, this two-phase reaction was not dependent on mixing
rate for either isolated yield or purity. Step 1 was thus shown to
be robust.
Figure 1. X-ray structure of diethanolamine adduct 8.
ethanol in >80% recovery to improve chemical purity and
further upgrade the chiral purity. Crude 6 with purities as low
as 85% and 75% de could be readily upgraded to >99% and
>99.5% de by forming 8 and recrystallizing it from ethanol.
This breakthrough meant that the chiral purity of the final
product no longer required 1 of high de, and chromatography
was no longer necessary to upgrade chemical purity.
Compound 8 is stable at room temperature over several
years, unlike delanzomib, which requires storage at −20 °C. A
stockpile of 8 was prepared and then conveniently hydrolyzed
to delanzomib when needed to provide phase 2 supplies.
Use of 8 as a Prodrug. Both Velcade, a commercial
proteasome inhibitor, and delanzomib are administered by IV
infusion due to poor oral bioavailability.7 An oral dosage form
would be highly desirable both for the patient and the health
care professional administering the drug. The solid-state
stability of 8, combined with its rapid conversion to delanzomib
in simulated gastric juices, led to the investigation of 8 as a
The peptide coupling was also shown to be robust.
Substoichiometric quantities of reagents only impacted overall
yield, but not purity, with excess reagents being removed during
workup. A key correlation, however, was shown between the
Scheme 4. Phase 3 process
C
dx.doi.org/10.1021/op400010u | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
equivalents of DIPEA and optical purity. A minimum of 2.8
equiv of DIPEA was required to keep epimer formation below
the target of 1.0%.
and cooling to 20 °C. Product solids were collected by vacuum
filtration, washed with DI water (400 mL), and then dried (50
Torr, 60 °C) to yield 247 g (0.82 mol, 82%) of 7 with 99.0 A%
1
Through the use of DoE, the volume efficiency of the
transesterification was improved by 40%. A combination of a
more concentrated reaction, along with higher amounts of
isobutyl boronic acid (3.5 vs 2.5 equiv), resulted in a shorter
reaction time and allowed the equilibrium to be shifted further
towards completion. Surprisingly, mixing was shown to have no
impact on this two-phase system other than on reaction time.
Higher than stoichiometric quantities of diethanolamine in the
adduct formation led to lower yields, presumably due to an
increase in polarity of the system, resulting in higher solubility.
The DEA charge is now calculated on the basis of the purity of
the crude delanzomib following the transesterification. Higher
recoveries of adduct have thus resulted.
purity. H NMR (400 MHz, DMSO-d6) 12.9 (s, 1H, b), 8.71
(d, J = 9.16 Hz, 1H), 8.23 (d, J = 7.24 Hz, 1H), 8.1 (m, 3H),
8.03 (d, J = 7.0 Hz, 1H), 7.55 (m, 3H), 5.34 (s, 1H), 4.46 (dd, J
= 2.52, 9.16 Hz, 1H), 4.34 (dd, J = 1.92, 6.24 Hz, 1H), 1.15 (d,
J = 6.4 Hz, 3H).
N-[(1S,2R)-1-[[[(1R)-1−1[(3aS,4S,6S,7aR)-hexahydro-
3a,5,5-trimethyl-4,6-methano-1,3,2-benzodioxaborol-2-
yl]-3-methylbutyl]amino]carbonyl]-2-hydroxypropyl]-6-
phenyl-2-pyridinecarboxamide (5). A mixture of 7 (239 g,
0.79 mol), HATU (319.6 g, 0.84 mol), 1 (238.1 g, 0.79 mol),
and DMF (1.2 L) was cooled to 0 °C, where DIPEA (420 mL,
311.6 g, 2.4 mol) was added dropwise over an hour. After
stirring at 0 °C for 2 h, the reaction was quenched into 1:1 DI
water/EtOAc (4.8 L). The layers were separated, and the
organic layer was washed with 10% aq NaH2PO4 (1 × 2.4 L),
8% aq NaHCO3 (2 × 1.5 L), and brine (1 × 2.4 L), before
drying over Na2SO4 (240 g), filtering, and concentrating to
dryness in vacuo (40 mbar, 40 °C) to a light-brown foam
weighing 493 g (435 g corrected, 99.6%) with a purity of 90.6 A
%. 1H NMR (400 MHz, DMSO-d6) 8.98 (d, J = 2.99 Hz, 1H),
8.76 (d, J = 8.55 Hz, 1H), 8.2 (m, 3H), 8.11 (t, J = 7.71 Hz,
1H), 8.02 (d, J = 7.54 Hz, 1H), 7.54 (m, 3H), 5.26 (d, J = 4.95
Hz, 1H), 4.49 (dd, J = 4.22, 8.52 Hz, 1H), 4.13 (m, 2H), 2.6
(m, b, 1H), 2.19 (m, b, 1H), 2.02 (br m, 1H), 1.83 (t, J = 5.38
Hz, 1H), 1.75 (br s, 1H), 1.68 (br m, 1H), 1.62 (d, J = 13.9 Hz,
1H), 1.36 (d, J = 10.05 Hz, 1H), 1.3(br m, 3H), 1.22 (d, J =
11.65 Hz, 6H), 1.12 (d, J = 6.26 Hz, 3H), 0.84 (d, J = 6.57 Hz,
6H), 0.79 (s, 3H).
6-(2S,3R)-N-[(1R)-1-(1,3,6,2-dioxoazaborocan-2-yl)-3-
methylbutyl]-3-hydroxy-2-[(6-phenylpyridin-2-yl)-
formamido]butanamide (8). A mixture of 5 (490 g, 0.79
mol), MeOH (2.35 L), heptanes (4.35 L), (2-methylpropyl)-
boronic acid (283 g, 2.78 mol), and a solution of 37% HCl (65
mL) in DI water (300 mL) was stirred at 20 °C for 20 h. The
layers were separated, and the lower MeOH phase was washed
with heptanes (1 × 4.35 L) before concentrating in vacuo (54
Torr, 38 °C). The resulting aqueous slurry was dissolved in
EtOAc (4.35 L), to which was slowly added 8% aq NaHCO3
(3.5 L). The layers were separated, and the EtOAc was washed
with 8% NaHCO3 (1 × 3.5 L) and brine (1 × 2.5 L) before
being assayed for product purity (94.2%) and transferred to a
clean reactor. Diethanolamine (78.7 g, 74.9 mol, 1.0 equiv) was
charged, and the resultant slurry was stirred at 20 °C for 12−18
h, before the white solid was collected by vacuum filtration in a
sealed filter. The wet cake was washed with EtOAC (1.3 L),
transferred to an isolator, and dried (50 Torr, 50 °C) to yield
289 g (0.6 mol, 88%) of 8 with a purity of 97.1 A%. This crude
product was recrystallized from 7 vol of absolute ethanol to
yield 243 g (84.2%) with a purity of 99.8 A% and chiral purity
of >99.8% de. 1H NMR (400 MHz, DMSO-d6) 8.8 (d, J = 8.52
Hz, 1H), 8.2 (m, 3H), 8.1 (t, J = 7.68 Hz, 1H), 8.0 (dd, J = 6.7,
0.9 Hz, 1H), 7.5 (m, 3H), 7.2 (br d, 1H), 6.5 (br t, 1H), 5.1 (d,
J = 4.92 Hz, 1H), 4.5 (dd, 1H), 4.2 (m, 1H), 3.6 (m, 2H), 3.5
(m, 2H), 3.1 (m, 1H), 3.0 (m, 2H), 2.7 (m, 2H), 1.6 (m, 1H),
1.3 (m, 1H), 1.2 (m, 1H), 1.1 (d, J = 6.32 Hz, 3H), 0.8 (dd, J =
6.68, 6.53 Hz, 6H).
Finally, in the recrystallization of 8, a thorough solvent screen
determined that 7 vols of absolute ethanol was optimal. No
optimization was done on the final hydrolysis of 8 to
delanzomib, as this step was no longer necessary.
CONCLUSION
■
The process was improved by introducing a chromophore at
the beginning of the sequence to allow for easy process
monitoring and assay of intermediates, a deprotection step was
eliminated, volume and atom efficiency were improved, and
most importantly, a novel crystalline diethanolamine adduct
was discovered that both allowed for chemical and optical
purity upgrades and served as a novel prodrug. The overall yield
for the process was improved from 25% to 60%, with the
elimination of two chromatographic purifications and a
significant product quality improvement from 90−95% with
98% de to >99.5% with 99.8% de.
EXPERIMENTAL SECTION
■
1H NMR spectra were obtained using a Bruker 400 MHz
spectrometer in the solvents indicated. HPLC spectra were
collected on an Agilent 1100 series instrument using an Agilent
Zorbax XDB C-18, 4.5 mm × 150 mm column with 1 mL/min
flow in a gradient of ACN/10−90% water over 15 min. HPLC
analysis results are reported as area % (A%). Retention times: 4,
10.06 min; 5, 12.21 min; 7, 6.93 min; 8, 7.65 min. All reactions
were performed, unless otherwise specified, in a round-bottom
flask equipped with a thermocouple, nitrogen inlet, gas outlet,
condenser, and either magnetic or overhead stirrer.
(2S,3R)-3-Hydroxy-2-[oxo-2-(6-phenylpyridin-2-yl)-
ethyl]butyric Acid (7). A mixture of 100 g 6-phenyl-2-
pyridine carboxylic acid (100 g, 1.0 mol), toluene (1 L) and
thionyl chloride (147 mL, 240 g, 2.0 mol) was heated at 75 °C
for 20 h. After cooling to 20 °C, excess thionyl chloride was
removed by distillation on a rotary evaporator at 40 mbar and
50 °C, followed by a toluene chase (2 × 1 L). The resultant
solution was added to a solution of L-threonine (125.6 g, 1.05
mol) and sodium carbonate (234 g, 2.2 mol) in DI water (2 L)
at 10 °C. After warming to 20 °C over 30 min, the reaction was
stirred for 18 h before separating the layers. Methanol (1.2 L)
was added to the lower aqueous layer before acidifying to pH =
1 (12 N HCl as needed). After stirring for 1 h at 20 °C, the
product solids were collected by vacuum filtration, washed with
DI water then dried (50 Torr, 60 °C). The crude product was
combined with methanol (2 L), heated to 40 °C, and filtered to
remove salts before charging 2 L of DI water over 30−60 min
AUTHOR INFORMATION
Corresponding Author
*E-mail: Renee.Roemmele@TevaPharm.com
■
D
dx.doi.org/10.1021/op400010u | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Scott Field, Emily Kordwitz, Mark Olsen, Gregory
Gilmartin and Christopher Neville for analytical assistance. We
also thank Stuart Dodson for procurement support, Crispin
Cebula for kilo-lab support, James Haley for chemistry support,
and our management, Roger Bakale, Michael Kress and John
Mallamo.
REFERENCES
■
(1) Gallerani, E.; Zucchetti, M.; Brunelli, D.; Marangon, E.;
Noberasco, C.; Hess, D.; Delmonte, A.; Martinelli, G.; Bohm, S.;
Driessen, C.; et al. Eur. J. Cancer 2013, 49, 290−296.
(2) Chatterjee, S.; Iqbal, M.; Menta, E.; Oliva, A. Proteasome
Inhibitors and Methods of Using the Same. WO Pat. Appl. WO/2005/
016859, Feb. 2, 2005; Chem. Abstr. 2005, 142, 240569.
(3) (a) Matteson, D. S.; Majumdar, D. J. Am. Chem. Soc. 1980, 102,
7590−7591. (b) Matteson, D. S.; Sadhu, K. M. Organometallics 1984,
3, 614−618. (c) Matteson, D. S. Chem. Rev. 1989, 89, 1535−1551.
(d) Matteson, D. S.; Sadhu, K. M. U.S. Patent 4,525,309, June 25,
1985; Chem. Abstr. 1986, 104, 88877 (e) Pickersgill, I. F.; Bishop, J. E.;
Koellner, C.; Gomez, J.; Geiser, A.; Hett, R. U.S. Patent 7,714,159,
May 11, 2010; Chem. Abstr. 2005, 143, 387386.
(4) Montalbetti, C. A. G. N.; Falque, V. Tetrahedron 2005, 61,
10827−10852.
(5) Hall, D. G. Structure, Properties, and Preparation Of Boronic
Acid Derivatives. Overview of Their Reactions and Applications. In
Boronic Acids; Hall, D. G., Ed.; Wiley-VCH Verlag GmbH & Co.
KGaA: Weinheim, 2005; pp 14−15.
(6) (a) Reference 5; pp 55−61. (b) Gravel, M.; Thompson, K. A.;
́ ́
Zak, M.; Berube, C.; Hall, D. G. J. Org. Chem. 2002, 67, 3−15.
(c) Brown, H. C.; Gupta, A. K. J. Organomet. Chem. 1988, 341, 73−81.
(d) Mancilla, T.; Contreras, R. J. Organomet. Chem. 1986, 307, 1−6.
(7) Mateos, M.-V.; San Miguel, J. F. Ther. Adv. Hematol. 2012, 3,
117−124.
E
dx.doi.org/10.1021/op400010u | Org. Process Res. Dev. XXXX, XXX, XXX−XXX