Investigation of the mechanism of racemization of litronesib in aqueous solution: Unexpected base-catalyzed inversion of a fully substituted carbon chiral center

Article abstract of DOI:10.1002/jps.23918

Mitosis inhibitor (R)-litronesib (LY2523355) is a 1,3,4-thiadiazoline- bearing phenyl and N-(2-ethylamino)ethanesulfonamido-methyl substituents on tetrahedral C5. Chiral instability has been observed at pH 6 and above with the rate of racemization increas

Full text of DOI:10.1002/jps.23918

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Investigation of the Mechanism of Racemization of Litronesib in
Aqueous Solution: Unexpected Base-Catalyzed Inversion of a Fully
Substituted Carbon Chiral Center
STEVEN W. BAERTSCHI,1 PATRICK J. JANSEN,1 ROBERT M. MONTGOMERY,1 WILLIAM K. SMITH,1 JERRY R. DRAPER,1
DAVID P. MYERS,1 PETER G. HOUGHTON,1 V. SCOTT SHARP,1 ANDREA L. GUISBERT,1 HONG ZHUANG,1 MICHAEL A. WATKINS,1
GREGORY A. STEPHENSON,1 THOMAS M. HARRIS2
1Eli Lilly and Company, Lilly Research Laboratories, Indianapolis, Indiana 46285-3811
2Vanderbilt University, Department of Chemistry, Nashville, Tennessee 37235
Received 8 November 2013; revised 24 January 2014; accepted 10 February 2014
Published online 14 March 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23918
ABSTRACT: Mitosis inhibitor (R)-litronesib (LY2523355) is a 1,3,4-thiadiazoline, bearing phenyl and N-(2-ethylamino)ethanesulfonamido-
methyl substituents on tetrahedral C5. Chiral instability has been observed at pH 6 and above with the rate of racemization increasing with
pH. A positively charged trigonal intermediate is inferred from the fact that a p-methoxy substituent on the phenyl accelerated racemization,
whereas a p-trifluoromethyl substituent had the opposite effect. Racemization is proposed to occur through a relay mechanism involving
intramolecular deprotonation of the sulfonamide by the side chain amino group and attack of the sulfonamide anion on C5, cleaving
–
the C5 S bond, to form an aziridine; heterolytic dissociation of the aziridine yields an ylide. This pathway is supported by (1) a crystal
structure providing evidence for a hydrogen bond between the sulfonamide NH and the amino group, (2) effects of substituents on the
rate of racemization, and (3) computational studies. This racemization mechanism results from neighboring group effects in this densely
functionalized molecule. Of particular novelty is the involvement of the side-chain secondary amino group, which overcomes the weak
ꢀC
acidity of the sulfonamide by anchimeric assistance. 2014 Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm
Sci 103:2797–2808, 2014
Keywords: racemization; chiral inversion; mechanism; neighboring group; relay ionization; ylide
INTRODUCTION
to be transiently converted to a trigonal carbon by rupture
of the bond to one of its substituents. The chiral stability of
(R)-litronesib was excellent under acidic conditions but base-
catalyzed racemization was observed at pH values greater than
6. This observation was unexpected because reasonable acid-
catalyzed racemization pathways could be proposed, for exam-
Chiral integrity is essential to the biological function of many
drugs. The maintenance of chiral integrity during manufacture,
formulation, storage, and distribution is of paramount impor-
tance and is a critical aspect of the development of safe and
effective drugs.1–3 This paper describes an unexpected racem-
ization reaction observed with a novel drug candidate.
In recent years, the molecular motors known as kinesins
have emerged as potential targets for cancer chemotherapy.
Extensive screening has led to the identification of a num-
ber of cell-permeable small molecules that inhibit mitosis by
blocking the function of essential spindle proteins other than
tubulin. In particular, Eg5 plays a key role in mitosis includ-
ing chromosome positioning, separating the centrosome, and
establishing a bipolar spindle. One such Eg5 inhibitor is the
1,3,4-thiadiazoline (R)-litronesib (LY2523355, 1), which has un-
dergone clinical evaluation for therapeutic treatment of human
malignancies.
–
ple, cleavage of the C5 S bond by protonation of N3 to give car-
bocation (2), which would be stabilized by delocalization into
the attached nitrogen atom and phenyl ring (Scheme 1). On
the contrary, base-catalyzed processes that could be expected
to occur even at neutrality were more difficult to envision. As
a consequence, a mechanistic investigation was undertaken to
elucidate the operative pathway.
EXPERIMENTAL
Materials
Litronesib was obtained from Eli Lilly & Company. Other
reagents were obtained from Sigma–Aldrich Chemical Com-
pany and used as received.
During the development of a liquid formulation intended
for intravenous administration, the observation was made that
the chiral purity of (R)-litronesib fell during storage of aque-
ous solutions. Racemization requires the tetrahedral carbon
Nuclear Magnetic Resonance
The nuclear magnetic resonance (NMR) data were acquired
on a Varian VNMRS 400 spectrometer equipped with an auto-
mated triple band probe. For proton spectra, the typical acqui-
sition parameters included eight scans, a 4.5 s acquisition time,
a 45◦ pulse width, and a 4629.5 Hz sweep width. The FIDs were
zero-filled to 32K data points and multiplied by a matched filter
exponential window function prior to Fourier transformation.
Abbreviations used: DEA, diethyl amine; DMEA, dimethylethyl amine; IPA,
isopropyl alcohol; MP, mobile phase.
Correspondence to: Steven W. Baertschi (Telephone: +317-276-1388; E-mail:
baertschi@lilly.com)
This article contains supplementary material available from the authors upon
request or via the Internet at http://onlinelibrary.wiley.com/.
Journal of Pharmaceutical Sciences, Vol. 103, 2797–2808 (2014)
ꢀ
C
2014 Wiley Periodicals, Inc. and the American Pharmacists Association
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014
2797

2798
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Two different columns were used for the enantiomer analyses;
a 4.6 × 150 mm Daicel Chiralpak IA, 5 :m for the N-methyl
and N-acetyl derivatives of litronesib and a 4.6 × 150 mm Dai-
cel Chiracel OD-H, 5 :m for 1, and the p-trifluoromethyl and
p-methoxy analogs of 1. The columns were operated at 40◦C.
The flow rate was maintained at 5.0 mL/min. MP A was liquid
carbon dioxide and MP B was methanol containing 0.2% (v/v)
triethylamine. MP B was increased linearly from 5% to 15% at
1.25% per minute for 8 min. Chromatograms were extracted at
290 nm.
Preparative Chiral Separations
The preparative chiral separation of the racemic N-methyl
derivative of litronesib was carried out on an 8 × 35 cm Chi-
ralpak AD column, eluting with 15:85:0.2 ethanol–heptane–
DMEA (dimethylethyl amine) (400 mL/min, 290 nm UV detec-
tion). The resulting enantiomers were analyzed on a 0.46 ×
15 cm Chiralpak AD-H column, eluting with 15:85:0.2 EtOH–
heptane–DMEA at 0.6 mL/min (290 nm UV detection).
The racemic p-methoxy intermediate (18a) was separated
on an 8 × 35 cm Chiralpak AD column by eluting with 60/40
heptane–isopropyl alcohol (IPA) (400 mL/min, 310 nm UV de-
tection). The resulting enantiomers were analyzed on a 0.46 ×
15 cm Chiralpak AD-H column, eluting with 60/40 heptane–
IPA at 0.6 mL/min (280 nm UV detection).
The racemic p-trifluoromethyl intermediate (18b) was sep-
arated by preparative SFC using a Chiralpak AD-H column,
with CO2 and methanol containing 0.1% diethyl amine (DEA)
as cosolvent at a wavelength between 214 and 359 nm. The
resulting enantiomers were analyzed on a 0.46 × 15 cm Chi-
ralpak AD-H column, eluting with a gradient mixture of MP
A: 0.2% DEA in hexane and MP B: 0.2% DEA in EtOH at
0.8 mL/min (5%–15% B after 7 min, held for 8 min, lowered
back to 5% B in 1 min and then maintained for a further 9 min,
25 min total run time, 285 nm UV detection).
Scheme 1. Postulated acid-catalyzed racemization of (R)-litronesib.
This pathway was excluded by the effect of pH on the rate of racemiza-
tion.
The resulting spectra were baseline corrected and referenced
to the solvent signal (2.50 ppm). For carbon spectra, the typ-
ical acquisition parameters included a 2.6 s acquisition time,
a 45◦ pulse width, and a 25,000 Hz sweep width. The number
of scans varied according to the concentration of the sample.
The FIDs were zero-filled to 64K data points and multiplied
by exponential window function (2 Hz) prior to Fourier trans-
formation. The resulting spectra were baseline corrected and
referenced to the solvent signal (39.5 ppm).
Ionization Constants
Titrations were performed using Sirius Analytical T3
(T310020) with SiriusT3Control software and SiriusT3Refine
Version: 1.1.0.10. Three titrations were performed from pH
2 to 12, in a single vial at varying ratios of cosolvent
MeOH–water mixture; conversion to 100% aqueous pKa val-
ues was accomplished by extrapolation to 0% cosolvent via
a Yasuda–Shedlovsky plot. Percentages of cosolvent used for
FastpKaUV were approximately 25–34–46; for potentiometric,
30–40–50. Computational pKa predictions were made using
MarvinSketchTM 5.2.6 (2009) (ChemAxon, Ltd.).
Mass Spectrometry
High-resolution positive ion electrospray (ES+) spectra were
obtained by one of the two systems: (1) LC/MS using an Agilent
6230 time-of-flight (TOF) spectrometer coupled to an Agilent
1200 LC system. The system was controlled and data manipu-
lated with Agilent MassHunter software; (2) an LTQ Orbitrap
Discovery (ThermoScientific). Nominal mass ES+ spectra were
obtained using a Waters single quadrupole detector mass spec-
trometer with Waters Empower software.
HPLC Methods
X-Ray Crystallography
Achiral HPLC analyses were performed on a Waters Acquity A high-quality single crystal of litronesib was grown by re-
UPLC system with photodiode array and electrospray MS de- crystallization from 88% acetone and 12% isopropyl ether. The
tection. The column, a 2.1 × 100 mm BEH-C18, 1.7 :m (Wa- crystal was mounted on a fiber at 100(2) K. Data were collected
ters), was operated at 45◦C. The flow rate was maintained using a CuK" radiation source (8 = 1.54178 A) and a Bruker
˚
at 0.6 mL/min. Mobile phase (MP) A consisted of 95/5 (v/v) D8 based 3-circle goniometer diffractometer equipped with a
17 mM acetic acid titrated to pH 5.6 with ammonium hydrox- SMART 6000CCD area detector (Bruker-AXS, Madison, Wis-
ide/acetonitrile, and MP B was acetonitrile. MP B was increased consin). Cell refinement and data reduction were performed
linearly from 0% to 90% at 18% per minute for 5 min and held using the SAINT program V7.68a.4 The unit cell was indexed,
˚
at 90% for 1 min. Chromatograms were extracted at 240 nm.
having triclinic parameters of a = 6.2219(3) A, b = 9.8401(5)
◦
◦
˚
˚
Chiral supercritical fluid HPLC analyses were performed on A, c = 11.3183(6) A, and " = 96.676(3) , $ = 98.394(3) , ( =
an Agilent 1100 series HPLC modified to pump carbon dioxide 103.872(3)◦. The cell volume of crystal structure was 657.23(6)
3
3
˚
and equipped with an Aurora SFC Systems Fusion A5 module. A . The density was calculated to be 1.293 g/cm at 100 K. The
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014
DOI 10.1002/jps.23918

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
2799
Scheme 2. Preparation of 6 and 7.a
structure was solved by direct methods (SHELXS86). All non- Synthesis of Litronesib Derivatives 6 and 7 (Scheme 2)
hydrogen atomic parameters were independently refined. The
The p-methoxy derivative (6) of (R)-litronesib was prepared
in nine steps: (1) treatment of 1-(4-methoxyphenyl)ethanone
(13a) on a 55-g scale with bromine and AlCl3 in MTBE gave
bromo-ketone (14a) in 78% yield and 91% purity. (2) Amination
of 14a with urotropin and conc. HCl–ethanol in CH2Cl2 gave
amine hydrochloride 15a quantitatively; the material was a
single peak by HPLC. (3) Protection of amine 15a with Boc2O
gave Boc-amide 16a in 67% yield and 96% purity. (4) Reaction of
16a with thiosemicarbazide afforded thiosemicarbazone 17a in
84% yield and 99% purity. (5) Cyclization–acylation of 17a with
pivalyl chloride gave racemic thiadiazoline 18a in 68% isolated
yield and 97% purity. MS (ESI) m/z = 507.6 (M + H)+, calc:
506.66. 1H NMR (400 MHz, DMSO-d6) * 10.85 (s, 1H), 7.17 (d,
J = 8.8 Hz, 2H), 6.89 (d, J = 8.8 Hz, 2H), 6.71 (br s, 1H), 4.34 (dd,
J = 6.7, 14.48 Hz, 1H), 3.95 (dd, J = 5.9, 14.50 Hz, 1H), 3.74 (s,
3H), 1.37 (s, 9H), 1.28 (s, 9H), 1.17 (s, 9H). (6) Preparative HPLC
of 18a (5.2 g) on an 8 × 35 cm Chiralpak AD-H column eluted
with 60/40 (v/v) heptane–isopropyl alcohol at 400 mL/min gave
the individual enantiomers with ee > 99%. (7) The Boc group
was removed from (R)-18a by treatment of with 5M HCl–EtAc–
EtOH at room temperature to give hydrochloride (R)-19a in
quantitative yield and 100% purity. (8 and 9) Treatment of (R)-
19a with ClCH2CH2SO2Cl and Et3N in 2-MeTHF at −20◦C
gave sulfonamide (R)-20a that was treated in situ with EtNH2
to give p-methoxy derivative (R)-6 of litronesib in 60% yield and
98% purity. The (S)-enantiomer was prepared identically. MS
(ESI) m/z = 542.1 (M+H)+, calc: 541.24. 1H NMR (400 MHz,
DMSO-d6) * 7.23 (d, J = 8.7 Hz, 2H), 6.90 (d, J = 8.6 Hz, 2H),
4.28 (d, J = 13.7 Hz, 1H), 3.97 (d, J = 13.8 Hz, 1H), 3.74 (s,
3H), 3.21 (t, J = 6.3 Hz, 2H), 2.88 (t, J = 6.3 Hz, 2H), 2.53
(q, J = 7.1 Hz, 2H), 1.27 (s, 9H), 1.18 (s, 9H), 0.99 (t, J = 7.1
Hz, 3H). 13C NMR (101 MHz, DMSO-d6) * 177.4, 174.1, 158.3,
141.1, 134.7, 125.9, 113.7, 82.9, 55.1, 51.5, 46.6, 43.3, 43.0, 26.8,
26.5, 14.9.
space group choice of P1 was confirmed by successful conver-
gence of the full-matrix least-squares refinement on F2. The
final goodness of fit was 1.050. The final residual factor, for all
data, was R1 = 0.0447 and wR2 = 0.1078. The largest differ-
ence peak and hole after the final refinement cycle are 0.302
and −0.446 (e A−3), respectively. The absolute stereochemistry
(R enantiomer) was determined by refinement of the absolute
structure parameter to 0.055(18).
Computational Studies
All calculations were performed using the Gaussian 09 Rev
B.01 suite of programs6 with the M062X hybrid functional7 and
the 6–31+G(d) basis set. All geometry optimization and vibra-
tion frequency calculations were carried out in a water solvent
environment using the self-consistent reaction field model with
solvent cavity defined using the polarized continuum model and
the universal force field radii model. Internally stored param-
eters for water were used for solvent calculations. Vibrational
frequency analysis was performed to ensure the correct num-
ber of negative eigenvalues for each stationary point. Gibbs free
energies (ꢀG) are used throughout the discussion.
Thiadiazine 4
Litronesib (380 mg) was heated for 3 days at 70◦C in 200
mL of 50/50 (v/v) acetonitrile/100 mM (NH4)2HPO4. Ultra-
performance liquid chromatography analysis of the solution
indicated the presence of 4 at a level of 1.8%. A pure sample of
4 (3 mg) was isolated by preparative HPLC. HRMS (ESI-TOF)
m/z = 268.1127 (M + H)+, C12H18N3O2S+, 2.6 ppm error. 1H
NMR (500 MHz, DMSO-d6): * 8.58 (br s, 3H), 7.58 (d, 2H), 7.50
(m, 3H), 6.52 (br s, 1H), 5.59 (s, 1H), 3.81 (m, 2H), 3.39 (m, 2H),
3.35 (m, 2H), 0.74 (t, 3H). 13C NMR (125 MHz, DMSO-d6): *
158.5, 132.7, 129.2, 128.8, 128.5, 48.2, 46.0, 45.7, 42.7, 12.1.
DOI 10.1002/jps.23918
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014

2800
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
i
ii
iii
iv
v
vi
viii
vii
Scheme 3. Preparation of 10.a
The same procedure was used to prepare p-trifluoromethyl racemic amine salt 26 in 92% yield and 93% purity. (6 and 7)
derivative (R)-7. (1) Bromo-ketone 14b was prepared in 59% Acylation of 17 with ClCH2CH2SO2Cl/Et3N/2-Me-THF/−20◦C
yield and 95% purity on a 74-g scale. (2 and 3) Amination of generated the vinyl sulfonamide 27, which was treated in situ
14b and in situ protection gave Boc-protected amine 16b in with 70% aqueous ethylamine to afford the final racemic 10 in
100% crude yield and 58% purity. (4) Without purification, 16b 32% yield and 98% purity after chromatography. MS (ESI) m/z
1
was converted to thiosemicarbazone 17b in 86% yield and 94% = 526.2 (M+H)+, calc: 525.24 H NMR (600 MHz, CD3CN): *
purity. (5) Cyclization and acylation of 17b gave thiadiazoline 7.37 (m, 2H), 7.32 (m, 2H), 7.29 (m, 1H), 4.64 (d, J = 15.2 Hz,
18b in 57% isolated yield and 98% purity. MS (ESI) m/z = 545 1H), 4.36 (d, J = 15.2 Hz, 1H), 3.25 (m, 2H), 3.00 (m, 2H), 2.94
(M+H)+, calc: 544.63. 1H NMR (400 MHz, DMSO-d6) * 10.96 (s, 3H), 2.62 (q, J = 7.2 Hz, 2H), 1.32 (s, 9H), 1.22 (s, 9H), 1.15 (t,
(br s, 1H), 7.72 (d, J = 8.3 Hz, 2H), 7.48 (d, J = 8.2 Hz, 2H), J = 7.2 Hz). Long-range coupling (HMBC) from the methylene
6.89 (br s, 1H), 4.35 (dd, J = 6.9, 14.4 Hz, 1H), 4.00 (dd, J = 5.9, protons at 4.64, and 4.36 ppm to a methyl resonance at 37.0
14.41 Hz, 1H), 1.37 (s, 9H), 1.28 (s, 9H), 1.17 (s, 9H). (6) Prepar- ppm, and from the methyl protons at 2.62 to the methylene
ative SFC of 18b on a Chiralpak AD-H column eluted with CO2 carbon at 52.5 ppm, established the position of a single methyl
and MeOH/0.1% DEA gave the individual enantiomers with ee group on the sulfonamide nitrogen. (8) The enantiomers (R)-10
> 99%. (7) The Boc group was removed from (R)-18b to give and (S)-10 were separated by preparative chiral HPLC on an
hydrochloride (R)-19b in quantitative yield and 100% purity. 8 × 35 cm Chiralpak AD column eluted with 15:85:0.2 (v/v/v)
(8) Treatment of (R)-19b with ClCH2CH2SO2Cl gave sulfon- ethanol–heptane–DMEA at 400 mL/min and monitored at 290
amide (R)-20b that was treated in situ with EtNH2 to give nm. (R)-10, MS (ESI) m/z = 526.2 (M+H)+, calc: 525.24. 1H
p-trifluoromethyl derivative (R)-7 of litronesib in 60% overall NMR (400 MHz, DMSO-d6) * 7.37 (t, J = 7.60 Hz, 2H), 7.22–
yield and 98% purity. The (S)-enantiomer was prepared iden- 7.29 (m, 3H), 4.53 (d, J = 15.3 Hz, 1H), 4.32 (d, J = 15.3 Hz,
tically. MS (ESI) m/z = 580.1 (M+H)+, calc: 579.22. 1H NMR 1H), 3.33 (td, J = 7.2, 14.0 Hz, 1H), 3.27 (td, J = 7.0, 14.0 Hz,
(400 MHz, DMSO-d6) *7.73 (d, J = 8.3 Hz, 2H), 7.54 (d, J = 8.3 1H), 2.89 (t, J = 6.9 Hz, 2H), 2.85 (s, 3H), 2.55 (q, J = 7.2 Hz,
Hz, 2H), 4.29 (d, J = 13.6 Hz, 1H), 4.03 (d, J = 13.6 Hz, 1H), 2H), 1.30 (s, 9H), 1.18 (s, 10H), 1.00 (t, J = 7.1 Hz, 3H). 13C
3.23 (t, J = 6.3 Hz, 2H), 2.89 (t, J = 6.3 Hz, 2H), 2.53 (q, J = 7.1 NMR (101 MHz, DMSO-d6) * 177.5, 174.8, 142.7, 142.3, 128.6,
Hz, 2H), 1.28 (s, 9H), 1.18 (s, 9H), 0.99 (t, J = 7.1 Hz, 3H). 13
C
127.4, 124.1, 84.5, 51.9, 49.1, 43.0, 42.9, 40.5, 36.6, 26.6, 26.5,
NMR (101 MHz, DMSO-d6) * 177.6, 174.2, 147.2, 125.4, 82.0, 14.9.
51.5, 46.2, 43.3, 43.0, 26.6, 26.5, 14.9.
(S)-10, MS (ESI) m/z = 526.2 (M+H)+, calc: 525.24. 1H NMR
(400 MHz, DMSO-d6) * 7.36 (t, J = 7.6 Hz, 2H), 7.22–7.30 (m,
3H), 4.53 (d, J = 15.3 Hz, 1H), 4.32 (d, J = 15.4 Hz, 1H), 3.34
(td, J = 7.2, 14.0 Hz, 1H), 3.27 (td, J = 7.0, 14.0 Hz, 1H), 2.89
(t, J = 6.9 Hz, 2H), 2.85 (s, 3H), 2.55 (q, J = 7.1 Hz, 2H), 1.30 (s,
9H), 1.18 (s, 9H), 1.00 (t, J = 7.1 Hz, 3H). 13C NMR (101 MHz,
DMSO-d6) *177.5, 174.8, 142.7, 142.3, 128.6, 127.4, 124.1, 84.5,
51.9, 49.1, 43.0, 42.9, 40.5, 36.6, 26.6, 26.5, 14.9.
Synthesis of N-Methyl Derivative 10 (Scheme 3)
The (R)-enantiomer of 10 was prepared in eight steps from
2-bromo-1-phenylethanone (21): (1) treatment of 21 in acetoni-
trile with excess methylamine in methanol at low tempera-
ture (<0◦C) produced amine hydrobromide 22; (2) direct treat-
ment with (Boc)2O generated Boc-derivative 23 in 54% over-
all yield with a purity of 64% (GC-MS). The main impurities
were excess (Boc)2O and BocNHMe, but their presence did not
Synthesis N-Acetyl Derivative 11
affect the subsequent step. (3) Reaction of 23 with thiosemi- (R)-Litronesib (164 mg) in methylene chloride (15 mL) was
carbazide afforded thiosemicarbazone 24 in 40% yield and 97% treated with 0.5 mL of acetic anhydride (4.9 mmol) for 15
purity. (4) Cyclization and acylation of 24 with pivaloyl chloride min at ambient temperature. The reaction solution was washed
gave racemic Boc-protected thiadiazoline 25 in 78% yield and with pH 8 phosphate buffer, dried over anhydrous sodium sul-
95% purity. (5) Deprotection with 5M HCl gave the HCl salt of fate, and the solvent evaporated to give 11 as a viscous oil;
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014
DOI 10.1002/jps.23918

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Table 2. Chiral Stability of Litronesib as a Function of pHa
2801
purity > 99% by HPLC (220 nm). MS (ESI) m/z = 554.6
(M+H)+, m/z = 552.7 (M−H)−. 1H NMR (600 MHz, CD3CN):
* 9.20 (s, 1H), 7.34 (m, 4H), 7.28 (m, 1H), 6.14 (t, J = 7.0 Hz,
1H), 4.50 (dd, J = 7.4, 13.4 Hz, 1H), 3.96 (dd, J = 6.4, 13.4 Hz,
Stress Conditionsb
Time Stressed (h)
S (%)
pH 2.0, 5◦C
120
120
120
120
120
120
123
120
120
120
97
0.0
0.7
0.0
0.0
0.0
0.1
20.0
0.0
0.0
0.0
1H), 3.69 (t, J = 6.7 Hz, 2H), 3.34 (q, J = 7.1 Hz, 2H), 3.26 (t, pH 8.0, 5◦C
pH 2.0, RT
pH 3.0, RT
pH 4.0, RT
pH 6.0, RT
J = 6.7 Hz, 2H), 2.02 (s, 3H), 1.40 (s, 9H), 1.20 (s, 9H), 1.15 (t, J
= 7.1 Hz). 13C NMR (151 MHz, CD3CN): * 178.5, 176.4, 172.1,
143.6, 142.1, 129.6, 128.7, 125.6, 84.4, 50.6, 48.2, 44.1, 41.3,
40.1, 40.0, 27.1, 27.0, 21.6, 14.1. Long-range coupling (HMBC)
pH 8.0, RT
from the protons at 3.69, 3.34, and 2.02 ppm to a carbonyl res-
pH 2.1, 40◦C
onance at 171.1 ppm, along with the presence of NH signals at
9.2 and 6.14 ppm, established the location of the acetyl group.
pH 3.3, 40◦C
pH 4.2, 40◦C
pH 6.2, 40◦C
pH 8.4, 40◦C
1.8
40.1
96
aLitronesib concentration = 0.1 mg/mL.
RESULTS AND DISCUSSION
Discovery of Chiral Instability
b10 mM citrate; pH measured after addition of litronesib.
1.9
Stability studies of (R)-litronesib were carried out between pH
2.1 and 8.2 at 40◦C for up to 93 h using solutions containing
10 mM tartrate adjusted to the reported pH. No degradation
could be detected by HPLC analyses using achiral stationary
phases, but analyses using chiral stationary phases revealed
formation of the (S)-enantiomer occurring above pH 6, with
progressively increasing chiral instability at higher pH values;
additional studies were carried out with similar results in 10
mM sodium chloride, 50 mM phosphate, 10 mM sodium sulfate,
and 10 mM methanesulfonic acid (Table 1). A broad pH screen
was conducted in 10 mM citrate from pH 2 to 8.4 (Table 2).
Together, these data confirmed the chiral instability above pH
6, but no specific buffer or ion effects were detected. Chiral
instability was also observed in buffered acetonitrile–water so-
lutions (data not shown).
y = –0.0001x + 3.9129
R² = 0.6161
1.7
y = –0.0025x + 3.9025
1.5
R² = 0.9303
y = –0.0072x + 1.702
1.3
R² = 0.9955
5°C
20°C
40°C
1.1
0.9
0.7
0
20
40
60
80
100
The effect of temperature was investigated with reaction ki-
netics being measured at 5◦C, 20◦C, and 40◦C in 10 mM citrate;
the pH was adjusted to 8.4 after temperature adjustment and
addition of (R)-litronesib (Fig. 1). First-order rate constants
Hours
Figure 1. Plot of the log of [%(R) enantiomer-%(S) enantiomer] of
(R)-litronesib in pH 8.4 10 mM citrate at 5◦C, 20◦C, and 40◦C over
time.
Table 1. Chiral Stability of Litronesib at 40◦C
were calculated using the data from 0% to 40% racemization
[i.e., formation of up to 20% of the (S)-enantiomer].
Time Stressed
(h)
S Enantiomer
(%)
Conditionsa
Mechanistic Investigation
10 mM tartrate, pH 2.1
10 mM tartrate, pH 3.1
10 mM tartrate, pH 4.1
10 mM tartrate, pH 8.7
50 mM tartrate, pH 2.0
50 mM tartrate, pH 2.9
50 mM tartrate, pH 3.9
50 mM tartrate, pH 6.3
50 mM tartrate, pH 8.4
50 mM tartrate, pH 4.0
10 mM NaCl, pH 8.8
96
96
96
0.0
0.0
0.0
39.7
0.0
0.0
Evaluation of Achiral Degradation Pathways
As a first step in the mechanistic investigation, degradation
products arising during solution stress testing8,9 were reviewed
to see whether they offered any clue as to the racemization
process. HPLC analyses (achiral) of the degradation products
resulting from stress tests carried out on litronesib solutions
in pH 8 phosphate buffer (0.41 mg/mL, 24 h, 70◦C) indicated
11.1% degradation with formation of two major degradation
products. The dominant degradation product, ethylenesulfon-
amide 3, (9.4% by HPLC-UV peak area) was the result of a sim-
ple elimination of ethylamine from the parent molecule. The
minor product, thiadiazine 4 (1.1% by HPLC-UV peak area),
was of more interest because it arose by a complex process that
we viewed as potentially related to the mechanism of racem-
ization. The structures of both compounds were established
unequivocally by NMR spectroscopy. The formation of 4 can be
rationalized as a $-elimination process in which either the sul-
fur or the nitrogen attached to C5 is displaced; the proposed
124
120
120
120
121
120
97
120
120
96
96
96
96
121
120
0.0
1.3
44.8
0.0
42.9
44.1
44.1
0.0
0.0
0.0
1.4
33.8
10 mM Na2SO4, pH 8.6b
10 mM CH3SO3H, pH 8.6
50 mM phosphate, pH 2.2
50 mM phosphate, pH 3.3
50 mM phosphate, pH 5.0
50 mM phosphate, pH 6.4
50 mM phosphate, pH 8.3
aLitronesib concentration = 0.1 mg/mL.
bpH measured before addition of litronesib.
DOI 10.1002/jps.23918
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014

2802
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
a
b
Scheme 4. Achiral degradation pathways involving base-catalyzed $-elimination reactions. Pathway (A) yields degradation product (3). Pathway
(B) shows the proposed pathway to degradation product (4). The initial step was shown to be nonreversible by failure to observe deuterium uptake
in an H/D exchange experiment, eliminating Pathway B as a possible racemization mechanism.
–
process involving cleavage of the C5 S bond to give olefin 5 is racemization by stabilization of the negative charge, whereas
shown in Scheme 4. the methoxyl substituent would have the opposite effect. The
The possibility that racemization might follow a similar chiral stability of litronesib and the two analogs was evaluated
course involving formation of trigonal intermediate 5 (or the at pH 7.85 in 70/30 phosphate buffer/acetonitrile for 46 h at
analogous structure in which the C5 N bond had been bro- 40◦C (Table 3). This pathway had to be rejected because the
–
ken) followed by reversion to the thiadiazoline was evaluated results represented by the first-order rate constants shown in
by incubation of (R)-litronesib in CD3CN:D2O (95:5, v/v). After Table 3 failed to support the involvement of a benzyl anion;
24 h at 55◦C, chiral HPLC analysis of the litronesib revealed the p-trifluoromethyl substituent caused the rate to decrease
formation of 14% of the (S)-enantiomer (28% racemization), but by approximately a factor of three, whereas the p-methoxyl
NMR analysis indicated that deuterium had not been incorpo- substituent caused a greater than twofold increase.
rated into the diastereotopic protons of the methylene group of
the partially racemized litronesib. On the basis of this result,
Formation of a Cation at C5
the $-elimination racemization pathway was rejected.
As a pathway involving anionic species was not supported, con-
sideration was given to pathways involving cationic intermedi-
ates. Paths (a) and (b) in Scheme 6 can be proposed in which
Possible Formation of an Anion at C5
Consideration was given to the possibility that deprotonation of racemization would occur via intermediates in which the trig-
the pivalamide NH on C2 might initiate base-catalyzed racem- onal benzylic intermediate would bear a positive charge. The
–
ization by cleavage of the C5 N bond, which would lead to processes would be initiated by SN2 attack of the sulfonamide
benzyl anion 5 stabilized by both the phenyl and the diva- anion on the fully substituted carbon to form aziridines with in-
lent sulfur (Scheme 5). To test this proposal, analogs 6 and 7 version of configuration at C5. In Path (a), the departing group
of litronesib were prepared having electron-donating methoxyl would be the sulfur moiety to form aziridine 8a; in Path (b), the
and electron-withdrawing trifluoromethyl substituents, respec- departing group would be the nitrogen moiety to form aziri-
tively, on the para position of the phenyl ring. The trifluo- dine 8b. In each case, reversible opening of the aziridines to
romethyl substituent can be predicted to increase the rate of yield a benzyl cation and sulfonamide anion would provide a
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014
DOI 10.1002/jps.23918

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
2803
Scheme 5. Postulated base-catalyzed racemization pathway via benzyl anion 5. This pathway was excluded by the effect of phenyl substituents
on the rate of racemization.
Table 3. Racemization of Litronesib and Four Analogs (S, %)a
Time (h)
Litronesib (1)
p-Methoxy (6)
p-Trifluoromethyl (7)
N-Methyl (10)
N-Acetyl (11)
0
7
22
26
30
46
67
0.07
2.36
6.50
–
8.64
12.86
–
0.96
5.95
15.37
–
19.66
27.34
–
0.70
1.44
2.67
–
3.57
5.24
–
0.82
0.89
0.91
–
0.85
0.90
–
0.03
–
–
0.04
–
0.09
0.08
<0.00001
<0.01
93
–
–
–
–
kobs (h−1
)
0.0013
1.00
0.0029
2.23
0.0004
0.31
<0.00001
<0.01
Relative rate
aStudies were conducted at 40◦C in a 70/30 mixture of pH 7.85 buffer and acetonitrile.
work of Uda and Kubota12 is particularly instructive. They ob-
served that under neutral conditions the ring-chain equilib-
rium between acetone thiosemicarbazone and the correspond-
ing 2-amino-1,3,4-thiadiazoline lies essentially completely on
the side of the acyclic species. Likewise, for the thiosemicar-
bazones formed by condensation of acetone with 1-methyl- and
3-methylthiosemicarbazides, the thiadiazolines are below the
level of detection, but the thiosemicarbazone formed with
1,3-dimethylthiosemicarbazide exists totally as the thiadi-
azoline. Also, they observed that the thiosemicarbazone–
thiadiazoline equilibrium is shifted toward the thiadiazoline
by acid. Thus, the thiosemicarbazone and 1-methyl- and 3-
methylthiosemicarbazones of acetone exist as the thiadiazo-
lines in TFA but revert to the acyclic species on recovery from
the acidic solution. Basic conditions favor the acyclic species,
although 5,5-dimethyl-2-amino-1,3,4-thiadiazoline remains in
the cyclic form in sodium methoxide. We believe litronesib
racemization is an extension of these observations and rep-
resents the first example of 4-acylthiadiazoline ring-opening
occurring without concurrent loss of the acyl group.
pathway for racemization. Literature precedents suggest that
the phenyl and tosyl groups alone would not impart sufficient
stabilization of the positive and negative charges to promote
racemization by this route10–13; however, the additional charge
compensation in ylides 9a and 9b might lower the activation
energy sufficiently to promote racemization.
As a test of the validity of these pathways, ionization of
the sulfonamide was blocked by methylation. N-Methyl deriva-
tive 10 (Fig. 2) was prepared as a racemate by a de novo
synthetic pathway (see Experimental), the structure was con-
firmed by NMR using HMBC correlations to establish the lo-
cation of the methyl group, and the enantiomers were sepa-
rated by preparative chiral chromatography. Chiral stability
of the (R)-enantiomer was evaluated at pH 7.85 (70/30 phos-
phate buffer/acetonitrile) (Table 3). Less than 0.1% of the (S)-
enantiomer was detected after 46 h at 40◦C, whereas approx-
imately 13% of the (S)-enantiomer of litronesib was formed
under these conditions. Thus, methylation of the sulfonamide
nitrogen effectively shuts down the racemization process,
clearly demonstrating direct involvement of the sulfonamide.
Insight as to whether pathway (a) or (b) (Scheme 6) would
be preferred can be gained from consideration of the work of
others on thiosemicarbazones, which have been found to be
in a dynamic equilibrium with 1,3,4-thiadiazolidines.14–20 The
The synthetic route to 1,3,4-thiadiazolines provides fur-
ther insight into the racemization mechanism. Preparatively,
thiosemicarbazones of ketones are converted to the thiadiazo-
lines by treatment with acid chlorides or anhydrides; acylation
DOI 10.1002/jps.23918
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014

2804
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Scheme 6. Proposed relay mechanism for base-catalyzed racemization via an ylide. Both pathways are consistent with all experimental
observations, but path (a) is strongly favored by literature precedents and computational studies.
occurs at N4 (and also at the amino group on C2 if it does not al- pKa Measurements (see Fig. 2)
ready bear substituents).14,16 The stability of resulting acylated
The proposed ylide-mediated racemization pathway is initi-
ated by formation of the sulfonamide anion. Measurements of
the pKa values of the ionizable functional groups were there-
fore performed. The UV spectrum of litronesib contains a pH-
dependent 8max in the region of 300 nm (g > 10,000), attributed
to the thiadiazoline ring system. A pKa value of 10.15 for de-
protonation of the 2-NH group on the thiadiazoline ring was
deduced by curve-fitting analysis from changes in the UV spec-
trum over the pH range from 8.9 to 11.9.
thiadiazolines is dependent on the 4-acyl group remaining in
place. Treatment of the N2,4-diacetyl derivative of 2-amino-5,5-
dimethyl-1,3,4-thiadiazoline with 1M NaOH, 1M HCl or even
with 10% NaHCO3 causes loss of the acetyl groups and ring
opening to yield acetone thiosemicarbazone.21 These observa-
tions about the general route for synthesis of thiadiazolines
reinforced by the equilibrium studies of Uda and Kubota16 lead
us to favor the C5-S cleavage over C5-N cleavage [pathway (a),
Scheme 6).
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014
DOI 10.1002/jps.23918

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
2805
Figure 2. pKa values observed for litronesib (1) and derivatives 10 and 11, and values calculated for model sulfonamide 12.
A potentiometric titration was then carried out to assess the
pKa values of the UV-silent sulfonamide and the secondary
amine. The titration of litronesib (0.197 mmol) with 0.0248M
NaOH showed two inflection points. The first was sharp per-
mitting a pKa of 8.04 to be deduced by curve fitting; the second
inflection was too broad to be analyzed in this manner but was
consistent with the pKa of 10.15 that has been assigned to the
2-NH by UV titration.
The observed pKa of 8.04 can be assigned to deprotonation
of the ammonium salt of the secondary amine. The pKa of
diethylamine is 11.09,22 but electron withdrawal by the sul-
fonamide reduces the basicity of the amino group in litrone-
sib. The value of 8.04 is consistent with the pKa of the amino
group of taurineamide, which has been reported to be 8.08.23
Junttila and Hormi24 have reported the pKa of methanesul-
fonamide to be 10.8. The alkyl substituent on the sulfonamide
in N-acetyl litronesib can be expected to raise the pKa of the
secondary amine to a value greater than 11. A pKa calcula-
tion carried out on model compound 12 using the computer
program MarvinSketchTM estimated values of 8.82 and 11.69,
Figure 3. Thermal ellipsoid drawing of the crystal structure of the
free base form of (R)-litronesib. The sulfonamide and amine nitrogen
atoms are 2.99A apart. The inferred hydrogen bond is indicated with a
respectively, for the secondary amino group and sulfonamide
NH (see Fig. 2). We conclude that the pKa of the sulfonamide
˚
group in litronesib is greater than that of the secondary amine dashed red line.
group. Thus, the order of ionization events occurring with in-
creasing pH is deprotonation of the secondary ammonium ion
followed by the C2 NH and finally the sulfonamide.
pKa value of approximately 10.0, which can be assigned to the
C2 NH, providing confirmation of the secondary amine assign-
ment for the 8.04 value that had been observed with litronesib.
By comparison, compound 10, having the sulfonamide blocked
Relay Ionization of the Sulfonamide by the Neighboring
Secondary Amine
In view of the fact that the first deprotonation event involves by methylation, had pKa values of 8.2 and 10.2. The chiral sta-
conversion of the ammonium ion to the free amine, the poten- bility of 11 was evaluated at pH 7.85 for 93 h at 40◦C with the
tial involvement of the amine in the racemization process was clear cut result that racemization was suppressed more than
investigated. The approach that was chosen involved acetyla- 100-fold (Table 3). This forces us to conclude that both the sul-
tion of litronesib to give 11 (Fig. 2), because the N-acetyl group fonamide and the secondary amine have obligatory roles in the
would eliminate both the basicity and the nucleophilicity of the racemization process.
amino group. The acetyl group was introduced by direct acety-
Our interpretation of this paradoxical result is that depro-
lation with acetic anhydride. The site of acetylation was estab- tonation of the sulfonamide NH is brought about intramolecu-
lished by NMR spectroscopy, which showed the expected de- larly by proton transfer to the secondary amine rather than by
shielding of the adjacent methylene protons along with HMBC direct reaction with hydroxide ion. An X-ray crystal structure
correlations between the methylene protons and the carbonyl of the free base form of litronesib (Fig. 3) shows that the two
˚
carbon of the acetyl group. UV titration of 11 showed only a nitrogen atoms lie close enough together (2.99 A) to support a
DOI 10.1002/jps.23918
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014

2806
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
involve dissociation of the aziridine to form ylide 9a, although
the aziridine was not detected during the course of our investi-
gations of the racemization mechanism.
Computational studies argue for the intermediacy of
aziridines and for a carbon–sulfur cleavage in the racemiza-
tion process as shown in Scheme 6, pathway (a). Density
functional theory calculations of the potential energy surface
carried out at the M062X/6–31+G(d) level of theory for the
carbon–sulfur cleavage depicted as pathway (a) in Scheme 6
showed well-defined wells for both thiomethine ylide 9a
and the aziridines 8a. The reaction coordinate is shown in
Figure 5. Simultaneous with evaluation of the energetics of
the direct and stepwise pathways, a comparison was made be-
tween pathways (a) and (b) in Scheme 6. Potential energy min-
ima were found for aziridines (S) and (R)-8b, although they
had significantly higher energies than the aziridines in path
(a). Collectively, these calculations provide strong support for
racemization occurring via aziridines and that pathway (a) in
Scheme 6 involving aziridines 8a and ylide 9a is the likely
route. The 16.3 kcal/mol energy barrier difference between the
neutral form of litronesib and the aziridine form 8a coupled
with the 5.8 kcal/mol energy barrier between the aziridine form
and the zwitterionic species 9a indicates that the steady-state
Figure 4. Six-membered ring created by the hydrogen bond between
the sulfonamide and the secondary amine.
hydrogen bond between them. Although the X-ray structure did
not show proton locations, the orientation of the sulfonamide
–
–
N S and N C bonds permits the sulfonamide proton location
to be calculated and placed in juxtaposition with the amino
group. The inferred hydrogen bond is shown in Figure 3 as a
red dashed line. Figure 4 shows the six-membered ring created
by the hydrogen bond. The ring is in a chair conformation with
the configurations of the ethyl substituent and thiadiazoline
substituents equatorial and axial, respectively.
We conclude the racemization process is being initiated by
deprotonation of the secondary ammonium ion. The resulting
amine deprotonates the sulfonamide, which in turn attacks C5
to yield aziridine 8a. The actual racemization process must
50
42.5
9b
40
30
26.1
(S)-8b
25.8
(R)-8b
20
18.8
9a
13.0
(S)-8a
13.0
(R)-8a
10
R-Litronesib
0
(R)-1
0.0
–0.5
–0.5
–3.3
–10
S-Litronesib
(S)-1
S-Litronesib
(S)-1
Reaction coordinate
–
–
Figure 5. Potential energy surface for the rearrangement of (R)-litronesib to (S)-litronesib via either a C S or C N bond cleavage pathway.
Relative free energy of stationary points is normalized to activated (R)-litronesib (R)-1. The activation step (intramolecular deprotonation) is
shown to require 3.3 kcal/mol. The computational results indicate the kinetically favored pathway to racemization of (R)-litronesib (R)-1 to
(S)-litronesib (S)-1 is through the C-S cleavage pathway.
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014
DOI 10.1002/jps.23918

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
2807
concentrations of the aziridines will be too low to detect by Explorer, Shanghai, China) for the synthesis of 6 and 7; Gejian
HPLC or the usual spectroscopic tools. Fundamentally, racem- Yang (PharmExplorer) for synthesis of racemic 10; Eric Seest
ization is a slow process, although sufficiently rapid to be a (Lilly) for separation of the enantiomers; Jinxing Ye, Feng Yu,
concern when needing to maintain the enantiopurity of a drug Haoxiang Hu, and Junzhu Lu (East China University of Sci-
during long-term storage in a liquid formulation.
ence and Technology, Shanghai, China) for synthesis of racemic
17; and Matt Belvo (Lilly) for separation of the enantiomers.
The authors declare no competing financial interest.
CONCLUSIONS
(R)-Litronesib (LY2523355) was discovered to have chiral in-
stability under neutral to basic conditions. An investigation
of the mechanism of this instability revealed an unexpected
base-catalyzed pathway with neighboring group participation
of the secondary amine with the sulfonamide. In support of
this pathway, racemization experiments were conducted with
four analogs of litronesib. Involvement of charge on the phenyl
REFERENCES
1. Wozniak TJ, Bopp RJ, Jensen EC. 1991. Chiral drugs: An industrial
analytical perspective. J Pharm Biomed Anal 9:363–382.
2. De Camp WH. 1993. Chiral drugs: The FDA perspective on manu-
facturing and control. J Pharm Biomed Anal 11:1167–1172.
3. Lin G-Q, Zhang J-G, Cheng J-F. 2011. Overview of chirality and
chiral drugs. Chiral drugs. Hoboken, New Jersey: John Wiley & Sons.
group was probed with the p-methoxyphenyl (electron donat- Chapter 1, pp 3–28.
4. Sheldrick GM. 1990. Phase annealing in SHELX90: Direct methods
for larger structures. Acta Cryst A46:467–473.
ing) and p-trifluorophenyl (electron withdrawing) analogs. The
p-methoxy substituent accelerated racemization, whereas the
p-trifluoromethyl substituent had the opposite effect. These
results are consistent with racemization processes involving
a thiomethine ylide that places positive charge on the ben-
zylic carbon. Protection of either the sulfonamide nitrogen (as
the N-methyl derivative) or the secondary amine (as the N-
acetyl derivative) blocked racemization, indicating their oblig-
atory simultaneous involvement in the racemization process.
5. Sheldrick GM. 1993. SHELXS93. Program for crystal structure re-
finement. Go¨ttingen, Germany: Institute fur anorganische Chemie.
6. Frisch MJ, Trucks, GW, Schlegel HB, Scuseria GE, Robb MA,
Cheeseman JR, Scalmani G, Barone V, Mennucci B, Petersson GA,
Nakatsuji H, Caricato M, Li X, Hratchian HP, Izmaylov AF, Bloino
J, Zheng G, Sonnenberg JL, Hada M, Ehara M, Toyota K, Fukuda
R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H,
Vreven T, Montgomery Jr JA, Peralta JE, Ogliaro F, Bearpark M, Heyd
The preferred side-chain conformation revealed by X-ray crys- JJ, Brothers E, Kudin KN, Staroverov VN, Kobayashi R, Normand J,
Raghavachari K, Rendell A, Burant JC, Iyengar SS, Tomasi J, Cossi M,
Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V, Adamo C,
Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi
R, Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG,
Voth GA, Salvador P, Dannenberg JJ, Dapprich S, Daniels AD, Farkas
O, Foresman JB, Ortiz JV, Cioslowski J, Fox DJ. 2009. Gaussian 09,
Revision B.01. Wallingford, Connecticut: Gaussian, Inc.
7. Zhao Y, Truhlar DG. 2008. The M06 suite of density functionals for
main group thermochemistry, thermochemical kinetics, noncovalent in-
teractions, excited states, and transition elements: Two new function-
tallography is consistent with a hydrogen bond between the
sulfonamide NH and the free base form of the amine. A relay
ionization process is postulated in which the free base form
of the amine abstracts the sulfonamide proton and the result-
ing sulfonamide anion attacks C5 with expulsion of the sulfur
moiety to form an aziridine. Racemization is proposed to occur
by a reversible dissociation of the aziridine to form a thiome-
thine ylide, which derives stabilization from the partial nega-
tive charge on the hydrogen-bonded sulfonamide. Racemization
¨
–
–
might have involved cleavage of either the C5 S or the C5 N als and systematic testing of four M06-class functionals and 12 other
functionals. Theor Chem Account 120:215–241.
bond, but computational studies and literature precedents pro-
vide strong support for cleavage of the bond to sulfur.
8. International Conference on Harmonisation Q1A(R2). 2003. Guid-
ance for industry: Stability testing of new drug substances and prod-
ucts, (Second Revision).
9. 2011. Pharmaceutical stress testing: Predicting drug degradation.
In Drugs and the pharmaceutical sciences; Baertschi SW, Reed RA,
Alsante KM, Eds. Vol. 210. London, UK: Informa Healthcare.
10. Tseng CC, Terashima S, Yamada S-I. 1977. Stereochemical studies.
XLV. A novel regiospecific ring opening of optically active 2-substituted-
1-tosylaziridines. Chem Pharm Bull 25:166–177.
The racemization mechanism for litronesib is unique. Al-
though at the outset we expected racemization to involve acid
catalysis (see Scheme 1), the experimental results revealed
that the process was in fact base catalyzed. This extraordi-
nary mechanism results from neighboring group effects in this
densely functionalized molecule. Of particular novelty is the
involvement of the side-chain amino group in a relay ionization
process that overcomes the weak acidity of the sulfonamide.
11. Davis FA, Zhou P, Liang C-H, Reddy RE. 1995. Addition of dimethy-
loxosulfonium methylide to enentiomerically pure sulfinimines: Asym-
metric synthesis of 2-substituted aziridines. Tetrahedron Asymmetry
6:1511–1514.
ACKNOWLEDGMENTS
12. Ghorai MK, Das K, Kumar A, Ghosh K. 2005. An efficient route to
The authors gratefully acknowledge Peter Wipf (University of regioselective opening of N-tosylaziridines with zinc(II) halides. Tetra-
hedron Lett 46:4103–4106.
Pittsburgh), Nicholas A. Magnus (Lilly), and Marvin Hansen
(Lilly) for helpful discussions. We are grateful to the fol-
lowing people for important contributions to this research:
Benjamin A. Diseroad (Lilly) contributions to determining the
X-ray single-crystal structure of 1; Jaclyn Barrett (Lilly) for
determining the ionization constants; Jonas Buser and Tim
A. Smitka for NMR characterization of 4, 10, and 11; Scott A
Bradley and Shanna Neely for NMR characterization of 6, 7,
and 18; Thomas M. Zennie for performing the deuterium ex-
13. Colpaert F, Mangelinckx S, De Brabandere S, De Kimpe N. 2011.
Asymmetric synthesis of "-chloro-$-amino-N-sulfinyl imidates as chiral
building blocks. J Org Chem 76:2204–2213.
14. Karabatsos GJ, Vane FM, Taller RA, Hsi N. 1964. Structural stud-
ies by nuclear magnetic resonance. VIII. Ring-substituted phenylhy-
drazones, semicarbazones, and thiosemicarbazones. J Am Chem Soc
86:3351–3357.
15. Mayer KH, Lauerer D, Ring/Ketten-Tautomerie bei. 1970. 1.3.4-
Thiadiazolidine-thionen-(2) und ꢀ2–1,3,4-thiadiazolinen. Liebigs Ann
change NMR experiment; Yang Liu and Xianbo Wang (Pharm- Chem 731:142–151.
DOI 10.1002/jps.23918
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014

2808
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
16. Uda M, Kubota S. 1979. Ring-chain tautomerism of acetone N- 20. Murakata C, Kato K, Ohta Y, Nakai R, Yamashita Y, Takahashi T.
methylated thiosemicarbazones. J Heterocyclic Chem 16:1273–1275. 2008. Thiadiazoline derivative. Patent US7425636B2.
17. Kubota S, Toyooka K, Ikeda J, Yamamoto N, Shibuya M. 1983. 21. Kubota S, Ueda Y, Fujikane K, Toyooka K, Shibuya M. 1980.
Synthesis of 5-(substituted phenyl) 4-acetyl-2-methylthio-ꢀ2–1,3,4-
thiadiazolines and their oxidation with potassium permanganate.
J Chem Soc Perkin Trans I: 967–1971.
Synthesis of 4-acyl-2-(acylamino)-ꢀ2–1,3,4-thiadiazolines and 4-acyl-
2-amino-ꢀ2–1,3,4-thiadiazolines by acylation of thiosemicarbazones.
J Org Chem 45:1473–1477.
18. Tao EVP, Staten GS. 1984. Synthesis of some N-(4,5,5- 22. Perrin DD. 1965. Dissociation constants of organic bases in aqueous
trisubstituted-ꢀ2–1,3,4-thiadiazolin-2-yl)-N,Nꢁ-dimethylureas. Hete- solution. IUPAC Chem. Data Ser. London, UK: Butterworths.
rocyclic Chem 21:599–601.
23. Kirsch LE, Sihn Y-S. 1997. The effect of polyvinylpyrrolidine on the
19. Toyooka K, Takeuchi Y, Taira Z, Kubota S. 1989. Synthesis of novel
stability of taurolidine. Pharm Dev Technol 2:345–356.
chiral 2,3-dihydro-1,3,4-thiadiazoles with extremely high rotations: 24. Junttila MH, Hormi OO. 2009. Methanesulfonamide: A cosolvent
X-ray crystal structure of (2R)-D-mandelyl-5-methylthio-2-phenyl-2,3-
and general acid catalyst in Sharpless asymmetric dihydroxylations.
dihydro-1,3,4-thiadiazole. Heterocycles 29:1233–1236.
J Org Chem 7:3038–3047.
Baertschi et al., JOURNAL OF PHARMACEUTICAL SCIENCES 103:2797–2808, 2014
DOI 10.1002/jps.23918