Scale-up Synthesis of Tesirine

Article abstract of DOI:10.1021/acs.oprd.8b00205

This work describes the enabling synthesis of tesirine, a pyrrolobenzodiazepine antibody-drug conjugate drug-linker. Over the course of four synthetic campaigns, the discovery route was developed and scaled up to provide a robust manufacturing process. Early intermediates were produced on a kilogram scale and at high purity, without chromatography. Midstage reactions were optimized to minimize impurity formation. Late stage material was produced and purified using a small number of key high-pressure chromatography steps, ultimately resulting in a 169 g batch after 34 steps. At the time of writing, tesirine is the drug-linker component of eight antibody-drug conjugates in multiple clinical trials, four of them pivotal.

Full text of DOI:10.1021/acs.oprd.8b00205

Subscriber access provided by University of Sussex Library
Full Paper
Scale-up synthesis of tesirine
Arnaud Charles Tiberghien, Christina Louisa Von Bulow, Conor Barry, Huajun Ge, Christian
Noti, Florence Collet Leiris, Marc McCormick, Philip Wilson Howard, and Jeremy Stephen Parker
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.8b00205 • Publication Date (Web): 02 Aug 2018
Downloaded from http://pubs.acs.org on August 2, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Scale-up synthesis of tesirine
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
*
Arnaud C. Tiberghien Christina von Bulow , Conor Barry , Huajun Ge , Christian Noti ,
, †
†
†
§
∥
#
‡
†
Florence Collet Leiris , Marc McCormick , Philip W. Howard , Jeremy S. Parker
‡
†
Spirogen, QMB Innovation Centre, 42 New Road, E1 2AX London, U.K.
§
Pharmaron, No.6, Taihe Road, BDA,Beijing, 100176, P.R.China
∥Lonza AG, Rottenstrasse 6, CH - 3930 Visp, Switzerland
#
Novasep Ltd, 1 Rue Démocrite, 72000 Le Mans, France
‡
Early Chemical Development, Pharmaceutical Sciences, IMED Biotech Unit, AstraZeneca,
Macclesfield, UK.
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
TOC Graphic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
O
H
H
H
N
N
O
O
O
O
N
N
N
O
O
O
O
N
H
O
O
O
O
O
O
N
OH
H
N
O
O
O
O
N
H
B
O
O
SG3249, tesirine
A
N
O
O
O
N
H
H
C
N
O
N
O
O
SG3199
ACS Paragon Plus Environment
2

Page 3 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
ABSTRACT
This work describes the enabling synthesis of tesirine, a pyrrolobenzodiazepine antibody drug
conjugate drug-linker. Over the course of four synthetic campaigns, the discovery route was
developed and scaled-up to provide a robust manufacturing process. Early intermediates were
produced at kilogram scale and high purity, without chromatography. Mid stage reactions were
optimized to minimize impurity formation. Late stage material was produced and purified using
a small number of key high-pressure chromatography steps, ultimately resulting in a 169 g batch
after 34 steps. At the time of writing, tesirine is the drug-linker component of 8 antibody-drug
conjugates in multiple clinical trials, 4 of them pivotal.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
KEYWORDS
Pyrrolobenzodiazepine
Antibody Drug Conjugate
High pressure chromatography
Loncastuximab tesirine
Camidanlumab tesirine
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
INTRODUCTION
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
During the last 10 years, antibody-drug conjugates (ADCs) have made a welcome addition to our
arsenal in the fight against cancer. In 2011, Adcetris (Seattle Genetics) was approved for CD30
positive AML, followed by Kadcyla (Genentech/Roche) in 2013 to treat HER-2 positive breast
cancers. Last year, two further approvals (Mylotarg and Besponsa from Pfizer) validated the
ADC approach, where a potent anticancer agent is delivered specifically to an antigen-expressing
tumour target. The ADC field is now expanding rapidly, with more than 60 agents in clinical
1
-2
trials and many more in pre-clinical development.
An antibody-drug conjugate is typically represented as a three component system: the targeting
antibody, the linker, and the active drug (Figure 1a). Although it is desirable to optimize and tune
each and every one of these components, the linker and the drug can be treated as a single small
molecular entity: the drug-linker. In addition, antibody-drug conjugates are modular systems; the
same antibody may be used to deliver different classes of drugs. Similarly, the same drug-linker
may be conjugated to any appropriate tumor-targeting antibody.
In 2012, tesirine (SG3249) was developed by Spirogen, as a drug-linker combining a set of
desired properties: fast and straightforward conjugation to antibody cysteines by maleimide
Michael addition, good solubility in aqueous/DMSO (90/10) systems, and a traceless cleavable
linker system delivering the highly potent pyrrolobenzodiazepine (PBD) DNA crosslinker
3 4
SG3199 (Figure 1b,c).
ACS Paragon Plus Environment
4

Page 5 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1: A) Schematic depiction of an antibody-drug conjugate; B) tesirine, a drug-linker
featuring a dipeptide trigger; C) SG3199, the active drug released by tesirine cleavage in the
presence of peptidases.
Antibodies conjugated to tesirine have been shown to be highly efficacious in preclinical
5
-8
studies. As a result of the modular nature of antibody drug conjugates, and the wide interest for
efficacious drug-linkers, tesirine was licensed to a number of companies. At the time of writing,
tesirine was the drug-linker component of more than 15 clinical trials, either in solid or liquid
tumours (see Table 1), and there is a considerable requirement for robust processes providing
tesirine on scale.
ACS Paragon Plus Environment
5

Organic Process Research & Development
Page 6 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Sponsor
Indication
Target
Phase
Medimmune/AZ
Medimmune/AZ
Medimmune/AZ
Abbvie
MM/AML
Prostate
MM
ASCT2
PSMA
BCMA
DLL3
DLL3
DLL3
DLL3
DPEP3
DLL3
CD25
CD25
CD19
CD19
CD19
I
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
I
I
SCLC
II Trinity
Abbvie
SCLC
III Meru
Abbvie
SCLC
III Tahoe
Abbvie
NETS
I
Abbvie
Ovarian
Melanoma
Lymphoma
AML/ALL
B-NHL
DLBCL
B-ALL
I
Abbvie
I/II
I
ADCT
ADCT
I
ADCT
I
ADCT
II
I
ADCT
Table 1: Some examples of key tesirine antibody-drug conjugates clinical trials.
9
In 2016, we reported the synthesis of tesirine on discovery scale. Despite being modular and
robust, the synthesis was not scalable without optimization. Indeed, flash chromatography was
still used in the initial stages of the synthesis (where kilograms of intermediates are typically
required to supply a phase I study), and excessively so in the later stages. A few key reactions
ACS Paragon Plus Environment
6

Page 7 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
suffered from low yields, and other were the source of impurities difficult to separate from the
desired material.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2: Block synthesis of SG3249. Boxed intermediates were synthesized in facilities
equipped for high-potency cytotoxics. The longest linear sequence is 20 steps long.
In this article we describe the work carried out over four synthetic campaigns, to optimize this
synthetic route on scale. Chromatographic steps were eliminated as often as possible, and
especially in the early stages of the route. Synthetic processes were simplified, and often
telescoped. Individual yields were improved, and impurities levels were controlled. Analytical
methods for each intermediate and final product were optimized and performance tests were
conducted to ensure accurate monitoring of the reactions and determine the purity of the
products. Ultimately, these optimizations resulted in a synthesis with improved robustness and
manufacturability, a purity increase from an original 85% to 97%, and the production of a 169 g
batch. This batch size appears relatively small, but because of the potency of SG3199, the drug
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
delivered by SG3249, only very small amounts are enough to achieve a pharmacological effect.
To put this into context, a 169 g batch of SG3249 would be theoretically sufficient to provide
6
, 10
more than 100 000 doses of Lonca-T at the current clinical schedule (120 µg/kg q3w x2).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
RESULTS AND DISCUSSION
In the early stages of the synthesis (Aromatic A-ring, proline C-ring, and Val-Ala peptide
trigger), we sought to optimize the process for larger scale production. In particular, great efforts
went into producing pure material whilst eliminating all chromatography steps.
Nitro-aromatic acid A-ring 6
Scheme 1: Improved synthesis of TIPS-protected 6-nitrovanillic acid 6
Reagents and conditions: (a) K CO , NMP, BnBr, 60 °C, 93%; (b) AcOH, HNO , 22 °C, 75%;
2
3
3
(c) TFA, AcOH, 80 °C, 85%; (d) TIPSCl, triethylamine, THF, 10 °C, 100%; (e) Sulfamic acid,
NH OH, NaH PO NaClO , THF, water, 0 °C, 68%.
4
2
4,
2
ACS Paragon Plus Environment
8

Page 9 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Benzylvanillin 2 is commercially available, but it was found to be more economical to produce
it on site from vanillin and benzyl bromide. Originally, DMF was used as the solvent, but a
screen revealed NMP to be kinetically advantageous, with good conversions in 1h at 60 °C.
Isolation was straightforward after precipitation in water. Nitration of benzylvanillin was initially
performed at 12 °C, aiming to minimize by-product formations, and in particular, the ipso-
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1
nitrodeformylation product 7 (Scheme 2a). This side product is explained by the position of
1
2
the formyl para to the benzyloxy group. In 1996, Cotelle and Cateau postulated that ipso-
nitrodeformylation can occur if the formyl group occupies the most (or one of the most) electron-
rich position of the aromatic ring, thus explaining competition with nitration in position 6.
Scheme 2: a) Ipso-nitrodeformylation of benzylvanillin; b) Photodecomposition of 6-
nitrobenzyvanillin
The nitration of benzylvanillin was found to be exothermic, and concerns over thermal
accumulation on scale forced us to reassess the process. The order of addition was reversed
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
adding a solution of benzylvanillin to the nitrating mixture), and the operational temperature
was changed to 22 °C. At this temperature, nitration occurred almost immediately with very little
thermal accumulation, and with a manageable level of 7 (12% by HPLC). An reaction
calorimetry test (RC1) on 587 g of benzylvanillin (Figure 3) found heat generation to be easily
controlled by the rate of addition. The heat of the process was 85.35 kJ and the corresponding
molar enthalpy (ꢀHr) was 147 kJ/mol. These parameters, together with the onset decomposition
temperature of product 3 at 244 °C validated process safety for kg scale production. Precipitation
in water followed by recrystallisation in ethyl acetate afforded the 6-nitrobenzylvanillin 3 in
good purity (96.6%) and 75% yield. The product was protected from light, since o-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
nitrobenzaldehydes are known to photochemically convert to o-nitrosobenzoic acid (Scheme
13
2
b). Next, the benzyl ether was cleaved with TFA at 80 °C. The volume of TFA was kept to a
minimum (2 V) because of its cost and potential environmental impact. AcOH (1V) was added to
ensure efficient stirring. Phenol 4 was cleanly obtained by precipitation with heptane. With an
overall yield of 59% and a purity of 98.7%, the results for this 3 steps sequence were found in
14
line with the reports of Rakshit and co-workers.
On research scale, we had protected phenol 4 in a solvent-free reaction; effectively melting
TIPS-Cl and imidazole at 100 °C. Although this approach is attractive in terms of solvent
volume reduction, the process was difficult to control on scale, and the reaction was potentially
reversible. Instead, a base and solvent screen showed high conversions with triethylamine and
TIPS-Cl in DCM. Crude aldehyde 5 was used directly in the Pinnick oxidation to provide
carboxylic acid 6. Here, the chlorine scavenger was swapped from hydrogen peroxide, to
1
5
sulfamic acid, as described by Lindgren in his original conditions. This has many advantages
including improved safety (no oxygen production, scavenging of chlorine dioxide), and clean
ACS Paragon Plus Environment
10

Page 11 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
removal in the aqueous phase during work-up. After slurrying in heptane, pure A-ring 6 was thus
obtained in 40% yield over 5 steps, on kg scale, and avoiding any chromatography.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3: Reaction calorimetry test on benzylvanillin. 587 g of 2 (23.8 w/w %) was dosed into
the reaction mixture at 20 °C during 2 h. The mixture was stirred for an additional 1h at the same
temperature. Heat generation in pink. The sharp peak is caused by precipitation of the product.
ACS Paragon Plus Environment
11

Organic Process Research & Development
Page 12 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Pyrrolidine C-ring 14
Scheme 3: Improved synthesis of Pyrrolidine C-ring 14
Reagents and conditions: (a) K CO , MTBE, water, Cbz-Cl, 15 °C, 100%; (b) MeOH, DCM,
2
3
H SO , 40 °C, 81%; (c) THF, Water, LiCl, NaBH , 17 °C, 71%; (d) Toluene, triethylamine,
2
4
4
TBSCl, 25 °C, 100%; (e) IPA, 10% Pd/C, H , 30 °C, 61%.
2
The synthesis of pyrrolidine C-ring 14 was conducted in parallel. The route described by
16
Gregson and co-workers was adapted on scale (Scheme 3). Benzyloxycarbonyl protection of
trans-hydroxyproline 9 was achieved more advantageously in a MTBE/water system rather than
toluene/water, with a high purity grade of Cbz-Cl in 100% yield. Classical esterification
conditions employing methanol and catalytic sulfuric acid were found difficult to work-up on kg
scale. Instead a 5/1 mixture of DCM/methanol was used, allowing for a convenient aqueous
ACS Paragon Plus Environment
12

Page 13 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
work-up with NaHCO . Ester 11 was isolated in 81% yield (7.8 kg) after solvent evaporation.
3
The reduction of the methyl ester to alcohol 12 had been readily achieved with LiBH on
4
research scale. However, the quantities of reactive LiBH involved on kg scale posed a fire
4
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
hazard, thus substitution with the more stable NaBH was desirable. Trials with NaBH alone
4
4
showed that the reaction rate was much slower than with LiBH . Lithium chloride was added to
4
produce LiBH in situ thus conserving the initial reaction kinetics and low number of
4
1
7
equivalents. Next, achieving the selective silylation of a primary alcohol in the presence of
secondary alcohol 12 was found to be key to avoiding chromatography on scale. Historically,
this improvement of selectivity had been achieved by replacing imidazole with the bulkier DBU,
and a low number of equivalents of TBS-Cl (0.77 equiv); altogether with moderate success.
Chromatography was still required to remove the bis-silylated product, and recycle the starting
material. Here, further screenings revealed triethylamine to be a more appropriate base for this
1
8
transformation. The rate of the reaction was reduced, but higher selectivity ratios were obtained
i.e.: SM/Primary silylation/Bis-silylation 6/89/5). The chromatography stage was eliminated
(
and the crude 13 was used directly in the next step. After hydrogenolysis of benzyl carbamate 13,
the resulting amine was purified by precipitation as its oxalate salt. The impurities were removed
in the filtrate, and amine 14 was isolated as its free base in 61% yield over two steps after a basic
aqueous work-up. Altogether, 14 was synthesized in 35% over 5 steps, with high purity (99.5%),
on kg scale without chromatographic purification. It was later established that high purity was
necessary, both for A-ring 6 and C-ring 14, to achieve good yields in the subsequent amide
coupling.
ACS Paragon Plus Environment
13

Organic Process Research & Development
Page 14 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Branching A and C ring aniline 19
Scheme 4: Improved synthesis of key intermediate 19
Reagents and conditions: (a) EDCI, HOPO, DCM, 15 °C, 82%; (b) TEMPO, KBr, NaOCl,
NaHCO , 3 °C, 92%; (c) Tf O, 2,6-lutidine, toluene, -40 °C, (d) MeB(OH) ,
3
2
2
Pd(dppf)Cl .CH Cl , K PO , toluene, 65 °C, 44% (two steps); (e) Zn, AcOH, EtOH, water, 5 °C,
2
2
2
3
4
8
1%.
ACS Paragon Plus Environment
14

Page 15 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
9
The conditions for the coupling of 6 and 14 reported in the research paper (Scheme 4) suffered
from a number of issues; DCC could not be removed easily by work-up, and HOBt is classed as
a desensitized explosive, causing shipping restrictions and supply chain concerns. Instead, EDCI
1
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
and HOPO were used. Experiments conducted without HOBt or HOPO were substantially
lower yielding, thus demonstrating the activating properties of these agents. The high purity of
the starting materials, together with careful temperature control meant that very little impurities
were produced during the reaction. The coupled product 15 could be isolated in 82% yield and
9
9.8% purity after crystallization from ethanol/water, on a 5 kg scale. In the next step, the
secondary alcohol was oxidized to a ketone with TCCA/TEMPO. This combination is very
efficient, but can result in undesired chlorination of aromatic groups and alkenes. We
successfully substituted TCCA/TEMPO with DMP (89% on 4 kg batches). However, DMP is
costly, shock sensitive and the work-up can be challenging. With this in mind, the team selected
2
0
the simpler and cleaner Anelli-Montanari process (TEMPO/Bleach in buffered biphasic
conditions). Ketone 16 was obtained cleanly in 92% yield. Next, ketone 16 was transformed to
the thermodynamic enol triflate 17 (with the unsaturation conjugated with the nitrogen in
position 2,3). Although the conditions of this reaction were not dramatically altered, subtle
changes in the number of equivalents of triflic anhydride (1.5 equiv instead of 3 equiv), and 2,6-
lutidine (2 equiv instead of 4 equiv) meant that the amount of 2,6-lutidine triflate by-product was
considerably reduced. This by-product acted as a poison in the subsequent Suzuki coupling and
had to be controlled to a low level. The solvent was switched from DCM to toluene, allowing
2
3
telescoping of the triflation with the next step. Introduction of the methyl group in C2 by sp -sp
Suzuki coupling was particularly challenging. The operational temperature of 65 °C is very close
to the decomposition temperature of the triflate. Additionally, undesired reduction of 17 resulted
ACS Paragon Plus Environment
15

Organic Process Research & Development
Page 16 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
in triflate elimination to yield 20 (Scheme 5) which is difficult to separate from the product,
throughout the remainder of the synthesis, even by chromatography. A conditions screen looking
at the influence of bases, solvent, catalysts and temperature was conducted.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
Scheme 5: Impurity formation during the sp -sp Suzuki coupling
Pd(dppf)Cl .CH Cl
was found to be
a
superior catalyst to the original
2
2
2
bis(benzonitrile)palladium(II) chloride, Pd(PPh ) Cl , or the combination of palladium acetate
3
2
2
2
1
and RuPhos. In 1984, Hayashi and co-workers had suggested that the high activity of
Pd(dppf)Cl could be ascribed to the large P-Pd-P angle of the catalyst (99°). This, in turn, may
2
explain the improved selectivity and reduction in side-products such as 20 caused by β-hydride
22
elimination. Potassium phosphate remained the best base, but a lower number of equivalents
was used (3 equiv instead of 6 equiv) to avoid overloading the reactor with solid material. Toxic
triphenylarsine and solid silver oxide were eliminated, thereby considerably improving the work-
up. Other methyl donors such as trimethylboroxine or MeZnCl were explored with varying
degrees of success, but did not afford improved conditions. Finally, a chromatographic step
removed most impurities and controlled the level of by-product 20 down to 0.7%. As a result of
ACS Paragon Plus Environment
16

Page 17 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
these improved conditions, batches of C2-methyl product 18 could be produced in 44% yield and
over two steps, on kg scale. Next, the nitro functionality was reduced with Zn and AcOH to
provide aniline 19. Instead of pre-activating the zinc by washing it with dilute HCl, it was found
more convenient to activate the zinc in situ by adding 5% water in the solvent (ethanol). Under
these conditions, the reaction is rapid, even at low temperature (5 °C), and the exotherm can be
controlled by the rate of addition of 18 to the reducing mixture. Amine 19 was isolated in 81%
yield and 97% purity, (27% over 5 steps, 1 chromatography) and was used rapidly in the next
steps due to its relatively low stability.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Carbamate protection of Amine 19
Amine 19 occupies a key position in the synthesis as it is the starting material for two parallel
branches (Figure 2): the top branch with a simple alloc protection (Scheme 6), and the bottom
branch with the inclusion of the Val-Ala peptidic trigger. Both branches rejoin later in a
dimerization step to form 39. Amine 19 was therefore split in two batches. A split factor was
calculated based on subsequent yields and number of equivalents used in the dimerization.
Initially, this split ratio was fixed at 0.39/0.61 top branch/bottom branch. However, as the yields
improved and the number of equivalents in the dimerization step was optimized from 1.5 to 1.2
equiv, the split ratio changed to 0.58/0.42. It is likely that the ratio will be further optimized by
subsequent campaigns improving the process robustness, and providing ever more reliable yields
and equivalence factors. In any case, both branches follow the same chemistry, except for the
carbamate formation with Alloc-Val-Ala-PAB-OH 24.
ACS Paragon Plus Environment
17

Organic Process Research & Development
Page 18 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 6: Improved syntheses of carbamate protected PBD monomers 29 and 34
Reagents and conditions: (a) Allyl chloroformate, pyridine, DCM, -5 °C, 84%; (b) PTSA, THF,
Water, 35 °C, 87%; (c) Oxalyl chloride, DMSO, triethylamine, DCM, -70 °C, 81%; (d) TBS-
OTf, 2,6-lutidine, DCM, 25 °C, 76%; (e) LiOAc, DMF, Water, 25 °C, 76%; (f) Triphosgene, 24,
triethylamine, DCM, 25 °C, 69%; (g) PTSA, THF, Water, 25 °C, 68%; (h) Oxalyl chloride,
DMSO, triethylamine, DCM, -60 °C; (i) TBS-OTf, 2,6-lutidine, DCM, -15 °C; (j) LiOAc, DMF,
Water, 40 °C, 71% (3 steps and 1 chromatography).
ACS Paragon Plus Environment
18

Page 19 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
2
3
We have previously reported the synthesis of peptide trigger building block 24, following the
2
4
25
studies of Dubowchik and Jeffrey. In this work, we have optimized the synthesis to allow
production of highly pure material on 500 g scale (Scheme 7). Most notably, in step a, different
bases were screened to avoid formation of double alloc, or dimerisation impurities. The
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
combination of NaOH/ Na CO showed an improved purity profile. In step d, prolonged
2
3
slurrying with MTBE controlled the by-product level of EEDQ condensation (quinoline) to very
low levels. Compound 24 was obtained at a purity of 98.9%, in 48% yield over 4 steps and did
not require chromatography. The chiral purity was found to be 99.9% by chiral HPLC.
Scheme 7: Improved synthesis of Alloc-Val-Ala-PAB-OH (24)
Reagents and conditions: (a) Allyl chloroformate, NaOH, Na CO , water, MTBE, 20 °C, 97%;
2
3
(
b) HOSu, DCC, DCM, 20 °C, 92%; (c) L-alanine, Na CO , THF, water, 25 °C, 79% (d) 4-
2 3
aminobenzyl alcohol, EEDQ, THF, 25 °C, 68%.
Protection of amine 19 with allyl chloroformate and pyridine was straightforward, even on kg
scale, and did not require further improvements. However, the hydrolysis of primary TBS ether
ACS Paragon Plus Environment
19

Organic Process Research & Development
Page 20 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
5 originally relied on AcOH, in a mixture of water, THF and methanol. The removal of AcOH
on scale during the work-up was difficult. Instead, a method relying on low number of
equivalents of acid was investigated. Ultimately, 0.6 equiv of tosic acid (PTSA) in wet THF was
used, thus considerably simplifying the work-up. Flash chromatography purification was
performed here, in order to provide a high grade of material 26 going into the next reaction.
Oxidation of the primary alcohol gives an aldehyde which spontaneously ring-closes to form the
B-ring of the PBD system (step c, Scheme 6). The Swern reaction is well suited to this
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
26
transformation , and more so on manufacturing scale than on research scale. Indeed, the number
of equivalents of oxalyl chloride had to be precisely controlled, or impurities such over-oxidized
lactam and unreacted started material were observed. The best outcomes were obtained after a
number of trial reactions, narrowing down on the optimum number of equivalents of oxalyl
chloride. It goes without saying that the moisture content of the starting material was a key
parameter, and Karl-Fisher determination helped inform the team when selecting a range of
oxalyl chloride equivalences. Typically, a small excess was used to compensate for the
remaining moisture. In this case, with a 0.23% water content (0.07 equiv water), 1.03 equiv of
oxalyl chloride was used and the product 27 was found pure enough to be used as such in the
next step (starting material / product / over-oxidation, 0.5 / 94.3 / 3.9). Protection of the
secondary alcohol with TBS triflate and 2,6-lutidine did not require further optimization.
Similarly, cleavage of the phenolic silyl ether with lithium acetate in wet DMF, proved to be a
remarkably smooth, mild, high yielding and scalable method. A simple slurrying sequence in
hexane and ethyl acetate provided key PBD monomer 29 with 99% purity and 34% yield over 5
steps, with a single chromatographic purification.
ACS Paragon Plus Environment
20

Page 21 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
The second PBD monomer 34 was synthesized in an almost identical sequence, apart from the
peptide trigger introduction by isocyanate chemistry. The chloroformate of benzylic alcohol 24
could not be used, since these types of compounds tend to eliminate CO and form their
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
chlorobenzyl analogues. On the other hand, aromatic aniline 19 is not nucleophilic enough to
easily react with activated carbonates (although it must be noted that Smith and co-workers
recently reported the clean condensation of an analogue of 19, at room temperature over a period
2
7
of 6 days, with a pentafluorophenyl carbonate ). For these reasons, the research route relied on
formation of an isocyanate intermediate by reaction of amine 19 with triphosgene (a solid, safer
alternative to phosgene gas). Several attempts were made to substitute triphosgene with low
toxicity reagents, but in these instances, were found inferior to the isocyanate and alcohol
condensation. A key improvement to this reaction was the simple solvent switch from
2
8
(
hygroscopic) THF to DCM. Moisture levels had to be stringently controlled or urea side-
products were observed as a result of isocyanate hydrolysis and self-condensation. Peptide 24
had a lower solubility in DCM than THF, but this was not found to be a limiting factor. These
improved conditions allowed the number of equivalents of 24 to be lowered from 1.5 to 1.05,
which in turn considerably simplified the work-up and chromatography. (An excess of 24
prevents silica gel chromatography by forming insoluble gel networks). Carbamate 30 was thus
obtained in 69% yield. TBS deprotection with PTSA revealed primary alcohol 31 as above.
Structure-activity relationship (SAR) of PBD species highlighted the importance to treat the
synthetic intermediates with caution from this point onward. Unless forming part of a pro-drug
strategy, ring-opened PBDs lacking an imine moiety (or equivalent carbinolamine and
29
aldehydes) are relatively non-toxic , and so are PBDs protected with biologically non-cleavable
carbamates such as Alloc. For example, the IC50 of alloc protected phenol 29 could not be
ACS Paragon Plus Environment
21

Organic Process Research & Development
Page 22 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
measured (>10 ꢀΜ, K562 CellTiter96 (MTS), 96 h incubation). But ring-closed intermediate 32
is protected by an enzymatically-cleavable dipeptide trigger, and could exert cytotoxicity in-vivo.
A change of strategy was therefore applied. The synthesis was completed in facilities equipped
to handle highly-potent compounds. Batch splitting and chromatography were permitted if
necessary, and high performance preparative chromatography was considered as a purification
option.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Primary alcohol 31 was ring-closed by Swern oxidation as described for compound 26. Again,
the oxalyl chloride equivalence was key, and in direct relation with the moisture content of the
starting material. For example, if thorough water azeotroping with dry toluene was conducted,
typical water levels would be at 0.06% w/w, and 1.03 equiv of oxalyl chloride were used. On the
other hand, when azeotroping with toluene was not used, and the water content was measured at
0
.2% w/w, 1.2 equiv of oxalyl chloride was used to fully consume the starting material and limit
the formation of over-oxidized impurity 36 (Scheme 9). The use of lower temperature could not
be considered to control over-oxidation due to lower solubility of the starting material in DCM.
Because of the cytotoxic properties of these PBD intermediates, isolation in solid form was not a
preferred option. Instead, the material was kept in solution (based on appropriate stability data in
solution) and used directly in the next step. Protection of secondary alcohol 32 with TBS-OTf
and 2,6-lutidine initially gave unacceptable levels of impurities. Mass analysis of the crude LC
profile showed up to 13% of TBS-OTf mediated carbamate cleavage product 35, presumably in a
3
0
reaction analogous to the BOC cleavage described by Sakaitani and Ofhune (Scheme 8).
Scheme 8: Postulated benzyl carbamate cleavage with TBS-OTf and 2,6-lutidine.
ACS Paragon Plus Environment
22

Page 23 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
A conditions screen looking at temperature and number of equivalents concluded that -15 °C was
low enough to avoid this side reaction, whilst allowing the protection to proceed (albeit slowly).
An excess of TBS-OTf (5 equiv) and 2,6-lutidine (6 equiv) was used to obtain full conversion in
2
0h. The solubilization of TBS-OTf in DCM proved to be useful for a better control of the
addition. Again, this step was telescoped with the next one to avoid any solid isolation. The
phenolic triisopropylsilyl ether was cleaved with lithium acetate in wet DMF as before, but it
was found that the reaction kinetics could be improved at 40°C without any degradation. The
extraction solvent was changed from ethyl acetate to Me-THF to avoid carrying traces of DMF
in the organic phase, which would have negatively impacted on the subsequent normal phase
chromatography. The removal of acid 37 (resulting from over-oxidation during the Swern
reaction, Scheme 9) was efficient with additional NaHCO wash. Remarkably, this three steps
3
sequence yielded pure phenol 34 after high pressure chromatography and precipitation in 71%
yield. This key intermediate was isolated as a solid.
Scheme 9: Postulated conversion of over-oxidation impurity 36 to acid 37.
ACS Paragon Plus Environment
23

Organic Process Research & Development
Page 24 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Next, the two phenolic monomers 29 and 34 were dimerized in two separate Williamson
etherification steps. In both steps, acetone was replaced with MEK to allow higher operating
temperatures, and a fine grade (typically ≤250 ꢁm) of potassium carbonate was used to improve
reaction kinetics. First, iodoalkane 38 was made by reacting phenol 29 with 5 equivalents of
diiodopentane. This high number of equivalents not only drives the reaction kinetics forward, but
also helps control the amount of homodimerization to below 10%. An aqueous work-up was
implemented to remove the reaction salts. Iodoalkane 38 was isolated by high pressure
chromatography and was stored in solution in ethyl acetate and used as such. This strategy
presents some advantages on scale due to the oily nature of 38 which prevents straightforward
transfers. In the second Williamson etherification, 38 and 34 were condensed under similar
conditions. An important factor here is the excess of iodoalkane 38 versus phenol 34. The
discovery route employed 1.5 equiv of 38 to fully consume 34 and ensure straightforward
chromatography. However, as was discussed above, this equivalence factor is taken into account
7 steps earlier to calculate the proportion of amine 19 to commit to each branch. When the yields
of the lower branch were dramatically improved, the amount of iodoalkane 38 became
unbalanced with the quantities of phenol 34 available. Fortunately, the dimerization was found to
work cleanly with a reduced equivalence of 38 (1.2 equiv), which partially restored the balance
ACS Paragon Plus Environment
24

Page 25 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
between the two branches of the route. A good conversion was obtained and crude dimer 39 was
taken directly to the next step without any purification.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 10: Final stages. Synthesis of tesirine (SG3249).
ACS Paragon Plus Environment
25

Organic Process Research & Development
Page 26 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Reagents and conditions: (a) Diiodopentane, K CO , MEK, 75 °C, 88%; (b) 34, K CO , MEK,
2
3
2
3
7
5 °C; (c) TBAF, AcOH, THF, 20 °C, 89% (2 steps); (d) Pd(PPh ) , pyrrolidine, DCM, 20 °C;
3 4
(
e) Mal-(PEG) -acid, EDCI, DCM/MeOH, 20 °C, 75% (2 steps).
8
We have previously described how unbuffered fluoride deprotection of the secondary TBS ethers
9
in 39 caused partial racemization of a key stereocenter in C11a. This base driven mechanism
can be prevented by buffering the mixture with a mild acid such as AcOH. Although this
represented a successful solution to the problem, we quickly noticed that the reaction kinetics
dropped in line with the pH – the reaction being unacceptably slow at acidic pH (< 4). This
ACS Paragon Plus Environment
26

Page 27 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
meant that a relatively narrow and optimum pH range (around 6.0) was controlled by the precise
equivalence of acid and TBAF. To add to our difficulties, and to our surprise, the actual fluoride
content of the TBAF solutions varied considerably between batches. These inconsistencies were
observed on multiple occasions, in different countries.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4: AcOH titration of a weak batch of TBAF
8.5
8
7
.5
7
6.5
6
5
.5
3
.9
4.1
4.3
4.5
4.7
4.9
5.1
5.3
5.5
TBAF equivalents
Note: Titration of 4 equiv of AcOH with a weak batch of TBAF in THF/water 80/20.
For example, when the reaction would not reach completion during one of the synthetic
campaigns, an acid-base titration (Figure 4) revealed that 5 “equiv” of this batch of TBAF were
required to neutralize 4 equiv of AcOH. The ratio used during the experiment was 4/ “3.9”
AcOH/TBAF which at this pH (5.5) prevented the reaction proceeding rapidly. The actual
concentration of the TBAF solution was found to be 0.72M (by anion-exchange
3
1
chromatography; see supporting information) instead of 1.0M, thus explaining these
inconsistencies. During a subsequent campaign, the reaction proceeded very rapidly, and partial
ACS Paragon Plus Environment
27

Organic Process Research & Development
Page 28 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
racemization was detected in the final step. This time, the concentration was calculated to be
1
.09 M instead of 1.0 M. On addition, fluoride precipitation can occur in bottles stored at 2 to 3
°C. With such wide error margin existing between real concentrations, and stated concentrations,
we introduced a strict quality control of the TBAF solutions, to determine their exact
concentrations before proceeding with the reaction (see supporting information for titration
method). In order to ensure the robustness of this critical step, a DoE study (supporting
information) was carried out to evaluate the impact of the main parameters (number of
equivalents of TBAF per TBS, and ratio TBAF/AcOH) on the conversion and diastereoisomer
content. The ratio TBAF/AcOH was the most important parameter to control the diastereoisomer
content. A low variability was found in the defined range 0.5-0.8, where the model was stable.
The amount of TBAF had a low impact on the diastereoisomer content. On the other hand, the
model was found to be linear for the conversion response. The amount of TBAF can be increased
to improve the kinetics of the reaction, while keeping a constant ratio of TBAF/AcOH. This DoE
screen revealed that the reaction kinetics could be improved by adding further TBAF/AcOH, and
that a margin of safety could be conserved by carrying out the reaction with a slight excess of
AcOH. The final conditions were 1.5 equiv of TBAF per TBS, and 0.63 equiv of TBAF per
AcOH. In these conditions, the reaction proceeded quite slowly (48 h), but guaranteed a low
level of racemization (<1%). High performance normal phase chromatography provided product
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
0 in 88% yield over two steps.
Next, the alloc carbamates were deprotected with tetrakis(triphenylphosphine)palladium(0) (0.02
equiv instead of 0.06 equiv) as described previously. Even with this reduced number of
equivalents, the reaction was rapid (30 min) and clean. A point of interest was the solubility limit
of the product in DCM, which became an issue upon scale-up. Relatively large volumes (27 V
ACS Paragon Plus Environment
28

Page 29 of 55
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
for the reaction, and 90 V for the extraction) had to be used to avoid losses upon precipitation.
To mitigate this issue, amine 41 was stored in a DCM / methanol mixture (95 / 5; 15 V) for
improved solubility and transfer purposes, and was used relatively rapidly. Prolonged solution
storage has been shown to favor the macrocyclic form of 41. Interestingly, this macrocycle
cannot be observed under the acidic aqueous HPLC conditions. However, it may be observed as
a streak on the TLC as it opens-up on the mildly acidic silica gel. (Figure S1, supporting
information). Understanding the different conditions controlling the equilibrium towards the
ring-opened (reactive) or ring-closed (unreactive) forms of 41 proved crucial to drive the amide
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
bond formation to completion in the final step. A moderate excess of EDCI and Mal-(PEG) -acid
8
(1.2 equiv) in DCM/MeOH (95/5) created a mildly acidic environment, which favored the
opening of the macrocycle and a full conversion. Many commercial batches of Mal-(PEG) -acid
8
were found to contain shorter Peg chains impurities such as Mal-(PEG) -acid which can impact
7
the final purity of SG3249. As a result, careful quality monitoring of the Mal-(PEG) -acid was
8
introduced to select the purest (99%) material. Crude SG3249 was purified by high performance
reverse phase chromatography, in mildly acidic conditions (0.01 % AcOH). These conditions
afforded a product of high purity, by removing any trace of starting materials and other
conjugable impurities. It is worth noting that more aggressive conditions (higher acid
concentration, heat) can trigger the aromatization of the PBD C-ring; an impurity with reduced
3
2
biological activity . Finally, high purity solid SG3249 was obtained after extraction and
concentration under vacuum (169 g, 75% yield over two steps, one chromatography).
ACS Paragon Plus Environment
29