Chromatography- and lyophilization-free synthesis of a peptide- linker conjugate

Article abstract of DOI:10.1021/op4002998

An optimized and scalable process to manufacture peptide-linker conjugate 1 is reported that avoids the chromatographic purification and lyophilization that are typically required for the isolation of this type of compound. An operationally simple protoco

Full text of DOI:10.1021/op4002998

Article
pubs.acs.org/OPRD
Chromatography- and Lyophilization-Free Synthesis of a Peptide-
Linker Conjugate
Javier Magano,*,† Brandon Bock,‡ John Brennan,† Douglas Farrand,‡ Michael Lovdahl,‡
Mark T. Maloney,† Durgesh Nadkarni,§ Wendy K. Oliver,‡ Mark J. Pozzo,∥ John J. Teixeira,† Jian Wang,‡
John Rizzo,⊥ and David Tumelty⊥
†Chemical Research & Development, Pfizer Worldwide Research & Development, Eastern Point Road, Groton, Connecticut 06340,
United States
‡Analytical Research & Development, Pfizer Worldwide Research & Development, Eastern Point Road, Groton, Connecticut 06340,
United States
§Biotherapeutics Pharmaceutical Sciences, Pfizer Worldwide Research & Development, 401 North Middletown Road, Pearl River,
New York 10965, United States
∥Biotherapeutics Pharmaceutical Sciences, Pfizer Worldwide Research & Development, 700 Chesterfield Parkway West, Chesterfield,
Missouri 63017, United States
⊥CovX Research, 9381 Judicial Drive, San Diego, California 92121, United States
S
* Supporting Information
ABSTRACT: An optimized and scalable process to manufacture peptide−linker conjugate 1 is reported that avoids the
chromatographic purification and lyophilization that are typically required for the isolation of this type of compound. An
operationally simple protocol has been developed that couples the peptide to the linker in DMF followed by precipitation with
MeCN. A scalable synthesis of the linker is also described which features the N-acylation of 2-azetidinone promoted by 1-
propanephosphonic acid anhydride (T3P). The number of operations during the second step of the synthesis (nitrobenzene
reduction to aniline) has been simplified by telescoping the aniline into the next step (reaction with diglycolic anhydride to form
an acid), thus avoiding an additional isolation. Finally, two efficient activation methods for the acid have been developed by
means of the corresponding pentafluorophenyl (PFP) and p-nitrophenyl (PNP) esters.
form drug substance 5 via azetidinone ring-opening and
creation of a β-alanine functionality.
INTRODUCTION
■
Biotechnology-based pharmaceuticals, or biopharmaceuticals,
have seen a considerable expansion in recent years.
“Bioconjugation”, which refers to the covalent derivatization
of biomolecules, can be employed to bring together two or
more molecules to form a complex that displays the combined
properties of each one of its individual components.1 This
technique has been applied to stabilize substances in blood, for
the protection of substances prone to enzymatic degradation,
and for decreasing the immunogenicity of polypeptides, among
others.2 An interesting application is in drug delivery systems,
such as in some oncology treatments, where the drug is
attached to a carrier by means of a linker or spacer that can
bind to its specific receptor contained on the cancer cell. As a
result, the therapeutic molecule can be directed to the desired
site of action.3 Both monoclonal antibodies4 and synthetic
polymers5 have been used as carriers.
Conjugate 1 (Figure 1) is an advanced intermediate in the
synthesis of antibody drug bioconjugate 5, currently under
evaluation at Pfizer as part of an oncology program.6 The
preparation of 5 involves two steps (Scheme 1): (a) activated
linker 2 undergoes reaction with a lysine residue on 22-mer
peptide 3, which is the therapeutic portion of the molecule, to
form peptide−linker conjugate 1 via a new amide bond; (b)
peptide−linker conjugate 1 is then subjected to a second
coupling with a lysine residue on monoclonal antibody 4 to
MEDICINAL CHEMISTRY SYNTHESIS OF 1
■
The original synthesis of 1 is shown in Scheme 2.6a,7 Acid 6 and
N-hydroxysuccinimide (7) underwent reaction with N,N′-
diisopropylcarbodiimide (DIC) as coupling reagent to afford
N-hydroxysuccinimido ester 8 in quantitative yield. Crude 8
was immediately used in the subsequent reaction with peptide 3
and N-methylmorpholine (NMM) as base in DMF to provide
conjugate 1, which was then subjected to extensive reverse-
phase chromatographic purification followed by lyophilization
to afford a 40% yield of purified 1. The lyophilized fractions
were reconstituted8 in MeCN/H2O 1:1 (v/v) to generate a
homogeneous lot of intermediate 1. After filtration of some
insoluble material,9 the filtrates were subjected to a second
lyophilization to generate 1 in 12% yield. This lengthy and both
energy- and solvent-intensive process was employed for the
generation of small batches of material (up to 50 g), but it
quickly became clear that its implementation to manufacture
even larger quantities of 1 would be impractical due to the very
low throughput. In addition, peptide 3 was only available at a
Received: October 20, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. Structure of peptide−linker conjugate 1.
Scheme 1. Preparation of peptide−linker−antibody bioconjugate 5
considerable expense from a commercial source,10 and the very
low overall yield would have made this method difficult to
implement on scale for cost reasons. When the need for larger
amounts of antibody drug bioconjugate 5 for clinical trials
arose, our group was requested to develop an optimized
synthesis of 1 capable of producing multihundred gram
quantities under current good manufacturing process
(cGMP) conditions in a cost-effective manner.11
to provide crude acid chloride 10. In a separate flask, 2-
azetidinone (11) was deprotonated at −70 °C with n-BuLi to
generate Li salt 12 which, without isolation, underwent reaction
with acid chloride 10 at 0 °C to give intermediate 13 in 91%
yield. The nitro group on 13 was reduced using catalytic
hydrogenation with 10% Pd/C to give aniline HCl salt 14·HCl
in 68% yield. The last step involved the reaction between 14
and diglycolic anhydride (15) using Hunig’s base to provide
acid 6 in 83% yield.
During preliminary experiments to determine the scalability
of this route, it was found that the yield for the coupling
between acid chloride 10 and azetidinone 11 was not
reproducible and dropped to 40−55% when the reaction was
run on 200-g scale. As a result, a search for alternative
PROCESS DEVELOPMENT FOR THE SYNTHESIS OF
LINKER 6
■
We first addressed the preparation of the linker portion of the
molecule. The Medicinal Chemistry synthesis of acid 6 is
shown in Scheme 3.12 Acid 9 was treated with Cl2SO at reflux
B
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 2. Medicinal Chemistry peptide−linker conjugation conditions
a
A major breakthrough took place when 1-propanephos-
phonic acid anhydride (T3P) was investigated.16 This coupling
reagent is commercially available as a 50 wt % solution in
EtOAc,17 and the reaction is usually carried out at ambient
temperature. Initial experiments showed that the addition of
T3P solution to a mixture of acid 9 and 2-azetidinone (11) in
DMF and pyridine as base gave desired product 13 in 20−45%
yield. A base (pyridine, DBU, DIPEA) and solvent (THF,
EtOAc, DMF, MeCN) screen revealed that consistent yields
around 40% could be obtained when the reaction was
performed with 3 equiv of DIPEA, 1.2 equiv of 2-azetidinone,
and 1.2 equiv of T3P in MeCN at 20 °C for 18 h. Further
optimization led to the use of 5 equiv of DIPEA, 1.5 equiv of 2-
azetidinone, and 1.5 equiv of T3P in MeCN at 20 °C for 20 h,
which provided coupling product 13 in 55−65% yield (Scheme
4). Clear advantages of this method are the operational
simplicity (slow addition of the 50 wt % T3P solution in EtOAc
to a mixture containing the two coupling partners and base in
MeCN), mild reaction conditions, and yield reproducibility.
Upon reaction completion, the solvent was removed under
vacuum, and the residue was redissolved in i-PrOAc. The
Scheme 3. Medicinal Chemistry synthesis of linker acid 6
a
Scheme 4. Optimized synthesis of linker acid 6
a
Reagents and conditions. (a) Cl2SO, reflux. (b) n-BuLi, −70 °C,
THF. (c) 0 °C, 91% (from 9). (d) H2, 10% Pd/C, MeOH, HCl, 40
°C, 68%. (e) DIPEA, CH2Cl2, rt, 83%.
conditions was carried out with the additional goal of avoiding
running the process at cryogenic temperature.
A literature search disclosed that the N-acylation of 2-
azetidinone has been accomplished through the use of acid
chlorides13 or anhydrides.14 After further exploring the first
approach, we noted that the preparation of acid chloride 10
using oxalyl chloride and the subsequent reaction with lithium
salt 12 prepared via deprotonation with LiHMDS at −30 to
−60 °C led to intermediate 13 but in low yield and purity.
Alternatively, the activation of acid 9 with CDI was readily
achieved, but the subsequent coupling with 2-azetidinone or its
anion failed in solvents such as CH2Cl2, THF, or EtOAc, most
likely due to the low nucleophilicity of this substrate. Finally,
KF on alumina was employed on the basis of some literature
reports15 to promote a slow deprotonation of the azetidinone,
but low conversion to 13 was observed over a 3-day period.
a
Reagents and conditions. (a) T3P (50% in EtOAc), DIPEA, MeCN,
20 °C, 20 h, 60−68%. (b) H2 (15 psig), 10% Pd/C (10 wt %), THF,
25 °C, 1 h. (c) THF, 22 °C, 15 min, 80% (2 steps).
C
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
organic phase was washed with aqueous citric acid, and after a
solvent switch to 2-propanol, intermediate 13 precipitated from
solution in high purity (≥98% area %). Before using this
material in the next nitro-reduction step, a Darco G-60
treatment in EtOAc was performed to remove some color
and trace impurities present in acid 9.18,19 A further benefit of
this carbon treatment was that the level of impurity 16 (Figure
2), resulting from the opening of the azetidinone ring by the
newly formed amino group in the subsequent nitro reduction
step, was kept below 0.3%.
solvent ratio of 9:1 i-PrOAc/THF was employed. The
telescoping of the THF solution of free base allowed for a
much simpler process and saved a considerable amount of time
in the plant.
Since this linker is also employed in another project that has
stricter residual solvent specifications (<0.25 wt % for any
individual solvent), an optimized crystallization protocol was
developed. Recrystallization of 6 from i-PrOAc at 70 °C (high
temperature was needed to dissolve the material in a reasonable
volume of solvent) caused product degradation, whereas
dissolution in acetone followed by solvent displacement with
i-PrOAc and crystallization provided material that was
otherwise pure but contained 0.5 wt % residual i-PrOAc.
Further experimentation provided a satisfactory recrystallization
method that called for dissolving acid 6 in THF (35 vol) at 30
°C and treating with Darco KBB (20 wt %) for 1 h (the carbon
treatment was included since the removal of colored impurities
appeared to lead to a more predictable crystallization behavior).
After filtering the activated carbon off, the filtrates were
concentrated to 10 vol, and i-PrOAc (20 vol) was added.
Concentration of this mixture at reduced pressure to 10 vol and
cooling to 0−5 °C afforded acid 6 in 80% recovery and >99%
chemical purity that met residual solvent specifications. This
procedure delivered significantly smaller particles (∼30 μm)
than the original THF/i-PrOAc crystallization (100−200 μm),
and the smaller particle size may explain why less solvent was
trapped in the crystals.
Figure 2. Impurities produced during the nitro reduction step.
The nitro reduction step was originally carried out in MeOH,
but due to the formation of considerable amounts of methyl
ester impurity 17 (Figure 2), the Medicinal Chemistry group
switched to dioxane to prevent its formation. Since this solvent
is a known carcinogen and undesirable on scale, we decided to
search for a safer alternative. Catalytic hydrogenation (15 psig)
with 10% Pd/C (10 wt %) in THF gave a fast reaction (<1 h),
but the formation of 1−5% of 16 occurred. Several additives
were tested to prevent or reduce the formation of 16 (HCl,
AcOH, water) but levels in the 1−2% range still occurred as
well as slower conversion. The presence of water was a concern
due to the potential opening of the azetidinone ring, but
running the reaction in a 9:1 v/v THF/H2O mixture only
caused a modest increase in impurity level. The best results
were obtained when starting material 13 was subjected to a
charcoal treatment as described above, which kept the amount
of impurity 16 at low levels (<0.3%) in the absence of additives.
Once full conversion of 13 to aniline 14 was attained, the
catalyst was filtered off, and the filtrates could then be treated
with 12 M HCl (1 equiv) to generate and isolate the
corresponding HCl salt, which is the preferred form for long-
term storage. However, during process optimization of the final
step to produce acid 6, we discovered that treating the THF
filtrates containing aniline 14 with 1 equiv of diglycolic
anhydride (15) led to the efficient formation of 6 without
the addition of base. The reaction was very fast (<15 min) in 60
volumes of solvent,20 and after concentration of the THF
solution, the addition of i-PrOAc and cooling caused acid 6 to
precipitate from solution in excellent yield (85−90%) and
purity (>98% area %). Unfortunately, this isolation method also
tended to result in relatively high residual solvent levels (0.5−
1% THF, 1−2% i-PrOAc), and an alternative solvent exchange
method became desirable to better control the final
crystallization. As a result, after reaction completion to form
6, the solution was concentrated and the product crystallized.
Subsequent slow addition of i-PrOAc to obtain a 3:1 ratio of i-
PrOAc/THF resulted in a yield of 85−90% and typical residual
solvent levels <0.2 and <0.5 wt % for THF and i-PrOAc,
respectively. No change in yield or quality was seen when a final
LINKER ACTIVATION
■
With acid 6 on hand, we turned our attention to the
conjugation step for the preparation of 1. As was shown in
Scheme 2, acid 6 was originally activated as its N-
hydroxysuccinimido ester prior to coupling with peptide 3.
We explored this approach by treating acid 6 with N-
hydroxysuccinimide (7) and N,N′-dicyclohexylcarbodiimide
(DCC) as coupling reagent to generate intermediate 8
(Scheme 5). Since the isolation of this material proved difficult
by crystallization, we decided to purify it by chromatography on
Scheme 5. First attempt at activation of linker 6 as N-
hydroxysuccinimido ester
D
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
silica, but signs of instability in contact with silica were observed
by TLC analysis. Even though 8 was isolated as a foamy solid in
an excellent 92% yield after chromatography, both HPLC and
1H NMR analyses showed a mixture of two major components.
One of them was identified as desired 8 but increasing amounts
of the second component were noticed after the solid had been
set aside for only a few hours. Further investigation resulted in
the identification of morpholine-3,5-dione 18 as the second
component by LC−MS, which formed via intramolecular
displacement of the N-hydroxysuccinimido group by the anilide
nitrogen on the molecule. This cyclic byproduct became the
major constituent of the isolated solid after only a few days at
ambient temperature. Since we envisioned this side reaction to
become even more problematic on scale due to the longer
timelines for processing and isolation, a search for an alternative
activating group for acid 6 was undertaken.
On the basis of the information obtained from experimenta-
tion related to the N-hydroxysuccinimido ester, it became clear
that a fine balance between reactivity toward the amino group
on the lysine residue of peptide 3 and stability to prevent
morpholine-3,5-dione 18 formation was necessary. We were
also looking for an activated ester that could be easily prepared
in high yield and stored over prolonged periods of time before
being used in the conjugation step. Finally, even though not
critical at this point, a low-cost activation method was desirable.
The preparation of the acid chloride derivative of 6 with either
Cl2SO or (COCl)2 or activation with reagents such as CDI or
chloroformate was ruled out since we assumed that, due to their
high reactivity, the resulting intermediates would readily cyclize
to give morpholine-3,5-dione 18. A literature search revealed
that esters derived from pentafluorophenol (PFP) have been
commonly employed for amide bond formation.21 Thus, when
acid 6 and pentafluorophenol (19) underwent reaction in the
presence of DCC in THF, ester 20 was obtained in 80−85%
yield (Scheme 6). Initially this intermediate was isolated via
CONJUGATION STEP PROCESS DEVELOPMENT
■
After having identified a satisfactory activated ester, we
investigated the final conjugation step to produce peptide−
linker 1. Our objective was the development of experimental
conditions that allowed for the isolation of 1 in high yield and
purity without resorting to the chromatographic purifications
and lyophilizations present in the original synthesis. If this goal
was accomplished, a much more cost-effective route to 1 would
be available for the production of large quantities of material.
The target specifications that had to be met for this product
were chemical purity ≥95% with no single impurity above 2%
and residual solvent content below 0.25 wt % for each
individual solvent.
Even though our group had developed the technology to
produce peptide 3 in-house,10 we employed high-quality
material (99% chemical purity by HPLC) for the work
described in this article that had been outsourced from a
vendor as the trifluoroacetic acid salt.22
A thorough screen of conditions was carried out that
encompassed solvents, bases, stoichiometry, reaction temper-
ature, concentration, and isolation method.
Because of the low solubility of 3, no conversion to 1 was
observed in solvents such as MeCN or MeOH. The addition of
water to these solvents to help bring 3 into solution led to low
conversions due to decomposition of activated linker 20 and to
the formation of thick gels after a few hours. As a consequence,
aprotic, polar solvents such as DMF, DMAc, NMP, and DMSO
were tested, which effectively dissolved 3 and allowed full
consumption of the peptide at 20 °C with 3 equiv of 20. The
cleanest reaction was obtained in DMF in combination with
NMM as base. Lower purities were obtained in DMAc, NMP,
and DMSO with bases such as TEA, DIPEA, or pyridine.
Conditions for the isolation of 1 were also investigated. The
simplest approach was the identification of a suitable
antisolvent that would precipitate 1 from solution followed
by filtration. A number of organic solvents were tested for this
purpose. Whereas toluene, EtOAc, THF, and MTBE gave
sticky, gel-like solids that were difficult to filter, MeCN
produced fine, free-flowing solids that were much more easily
handled. Filtration under nitrogen was mandatory at this point
to prevent moisture adsorption on the cake, which turned the
product into a gummy solid. The workup procedure involved
the slow transfer of the DMF solution of 1 after reaction
completion into MeCN to give a final 9:1 MeCN/DMF
mixture. After the resulting precipitate was aged for 1−2 h, the
solid was filtered and dried at 30−40 °C under vacuum.
With the base, solvent, and antisolvent choices already
identified, we looked at improving the impurity profile. The
reaction was very fast (<30 min) when run at 20 °C, and the
purities of the isolated material were in the 90−95% range. Up
to four major impurities were detected that caused a drop in
purity compared to the purity of the peptide 3 starting
material.23 These four byproducts were characterized as PFP
ester derivatives of 1 based on MS data (all of them displayed
the same mass), and when combined together, they amounted
to 7−8%. Interestingly, each impurity is a distinct monoester
corresponding to the four different carboxylic acid groups on
the peptide backbone, and no impurities with multiple esters
were detected. Therefore, it appears that once a PFP monoester
derivative forms, further esterification on that same molecule is
extremely slow. This fact is difficult to rationalize on the basis of
the existing physical separation between the acid groups, and at
Scheme 6. Synthesis of PFP-activated linker
chromatographic purification, but we eventually developed a
methodology that allowed for its crystallization from i-PrOAc.
Ester 20 is a fine, free-flowing solid that is stable for months at
room temperature in contact with air, without the need for any
special storage conditions and that shows no detectable
amounts of morpholine-3,5-dione 18. In addition, it displays
excellent reactivity in the final conjugation step to form 1 (vide
infra). Our group has scaled up the preparation of this
intermediate to produce 95 g of cGMP material.
E
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
this point, we do not have an explanation either for the
mechanism that leads to the formation of these impurities nor
for the observed chemoselectivity.24 On the basis of this
information, we investigated ways to minimize their formation
in order to meet the ≥95% purity specification for 1. A major
driving force for this was that any attempts to upgrade the
chemical purity of 1 by trituration or recrystallization had failed
due to the extremely low solubility of 1 and related byproducts.
We first looked at decreasing the reaction temperature. Thus,
we discovered that the conjugation between 3 and 20
proceeded even at −30 °C in DMF, but it was considerably
slower (18−20 h), necessitating up to 5 equiv of 20 to fully
consume the peptide. In addition, the longer reaction times
tended to give more byproduct formation as well as hazy
mixtures due to partial product precipitation. After further
experimentation, a compromise was found between the kinetics
of the process and a satisfactory impurity profile. The final
process that was implemented under cGMP conditions called
for running the reaction in 15 volumes of DMF (enough to
easily dissolve peptide 3 and prevent gelification25) with 3 equiv
of PFP ester 20 and 1.2 equiv of NMM at −15 to −18 °C. After
7 h, less than 1% of unreacted 3 remained, and the total PFP
ester byproduct level was kept at 1.5−2%. The mixture was
then filtered through a 0.45 μm in-line filter and slowly added
into 135 volumes of MeCN while using low agitation to
promote larger particle size and facilitate the subsequent
filtration. Solid precipitation occurred immediately, and after a
2-h aging period, the solids were filtered, washed with fresh
MeCN, and dried under vacuum. Unfortunately, GC headspace
analysis of the solid showed a DMF content of 4.5 wt %, most
likely due to the intentional slow agitation during the quick
precipitation of the product from solution and the resulting
entrapment of DMF. This issue was resolved by screening the
solid through a #20 hand-sieve followed by a MeCN reslurry
and subsequent drying at 40 °C to afford material in 83% yield
that met desired specifications in terms of purity and residual
solvent content. In retrospect, it is possible that a higher
agitation speed or subsurface addition of the product solution
into MeCN during the precipitation could have minimized the
amount of entrapped DMF, but this point was not
corroborated experimentally.
only 30 min, one of the three impurities was not detectable, and
the levels of the other two had decreased considerably. At the
same time, the overall purity had increased by about the same
amount. After 60 min, the remaining two impurities had almost
completely vanished, and the purity of 1 stayed unchanged.
This result seems to indicate that these impurities are short-
lived in this aqueous medium and revert back to the desired
product 1.
CONJUGATION OF P-NITROPHENYL ESTER OF 6
WITH PEPTIDE 3
■
Although the PFP ester impurities could be successfully
hydrolyzed in the histidine buffer solution prior to conjugation
with the antibody, we still needed to closely monitor their
hydrolysis by HPLC, and this proved to be rather challenging
from analytical and quality control perspectives. In addition, a
control experiment in which pentafluorophenol was added to a
solution of 1 that had been prepared via NHS ester activation
afforded a reaction mixture containing PFP ester impurities of
1. The facile formation of these PFP ester impurities prompted
us to evaluate the coupling of peptide 3 with a less reactive
phenyl ester derivative of 6 that would also lead to the
generation of a less reactive phenol coproduct to improve the
impurity profile. To test this hypothesis, we prepared p-
nitrophenyl (PNP) ester 22 in 68% isolated yield using
procedure similar to the one used for the synthesis of 20
(Scheme 7). The PNP ester was isolated as a pale-yellow solid
which, unlike NHS ester 8, was found to be stable at room
temperature for several days without any decomposition.
Scheme 7. Synthesis of PNP ester 22 for acid 6 activation
A final study was carried out to determine the stability of the
PFP ester impurities that were identified during the peptide−
linker conjugation reaction. We reasoned that these byproducts
would also couple to mAb 4 during the final process to prepare
the drug substance and might complicate the final purification
of 5. As a result, a sample of 1 with ∼2% PFP ester impurity
levels26 was placed in the same 1:1 20 mM histidine/propylene
glycol buffer at pH 6.5 that is employed to bioconjugate 1 with
4, and samples were pulled at regular intervals (Table 1). After
Compound 22 (3 equiv) was coupled with the free base of
peptide 3 in DMF at 0−5 °C for 8 h (Scheme 8). Desired
product 1 was isolated via precipitation with acetonitrile in 86%
yield. Analysis of the isolated product by HPLC and LC−MS
did not show the presence of any corresponding PNP ester
impurities originating from 1. Thus, PNP ester 22 satisfied the
delicate balance between long-term stability and adequate
reactivity without compromising product purity.
Table 1. PFP ester impurities time-dependent stability study
in 1:1 20 mM histidine/propylene glycol buffer at pH 6.5
time point
(min)
purity (1, impurity 1
impurity 2
(%)
impurity 3
(%)
entry
%)
(%)
CONCLUSIONS
■
1
2
3
4
5
0
30
95.6
96.5
96.5
96.6
96.7
0.75
0.53
n/d
0.68
0.14
0.05
n/d
We have developed an efficient conjugation protocol for the
coupling of a linker and a peptide as part of a program to
manufacture a peptide−linker−antibody bioconjugate for the
treatment of cancer. This novel approach allows for the
isolation of advanced intermediate linker−peptide 1 via direct
precipitation from a MeCN/DMF mixture followed by
0.26
0.09
60
n/d
90
n/d
n/d
n/d
120
n/d
n/d
n/d: not detected.
F
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 8. Coupling of PNP ester 22 with peptide 3
Table 2. Comparison between Medicinal Chemistry and cGMP processes for the conjugation of activated linker with peptide 3
process
activation method
chromatographies
lyophilizations
energy intensive
solvent intensive
overall yield (%)
Medicinal Chemistry
cGMP
NHS ester (unstable)
2
0
2
0
yes
no
yes
yes
12
PFP or PNP esters (stable)
83−86
filtration. The resource-intensive isolation protocol performed
originally and common to this class of compounds, which
involves extensive chromatographic purification and lyophiliza-
tion, has thus been avoided. One of the major improvements is
the dramatic increase in yield for this step, which has gone up
from 12% to 83%. Further advantages are the reduction in
solvent and energy consumption, which substantially lowers the
cost per gram and turns the process into a greener alternative.
An extensive screen of all the key parameters resulted in
reaction conditions that provided material of satisfactory quality
to meet stringent specifications in terms of purity and residual
solvent content. To the best of our knowledge, this method
represents an innovative approach toward the isolation of this
type of material that departs from the traditional chromato-
graphic purification. Table 2 summarizes the key features
described above for the Medicinal Chemistry and cGMP
processes.
We have also demonstrated an optimized route to the
activated linker that eliminated cryogenic conditions and the
use of n-BuLi, avoided the isolation of one of the intermediates
through telescoping, and identified the PFP ester as a suitable
substrate with the desired balance between stability for isolation
purposes and reactivity in the final conjugation step with the
peptide therapeutic agent. All these improvements have
translated into the manufacture of 91 g of compound 1
under cGMP conditions. In addition, the PNP ester derivative
has been shown to be a suitable alternative for acid 6 activation.
We believe that this technology has the potential to deliver
even larger quantities of material with little additional
optimization.
40; 23 min: 0/100; 25 min: 0/100; 25.1 min: 95/5; 30 min:
95/5. Flow rate: 1 mL/min.
Product purity for compound 1 prepared from 20 was
evaluated by HPLC using the following conditions: Column:
YMC-Pack ODS-A, 250 mm × 4.6 mm I.D., S-5 μm 12 nm,
P/N AA12S05-2546WT. Column temperature: 60 °C.
Detection: UV @ 220 nm. Sample dissolving solvent: 1:1
water/DMF. Mobile phase: A: 0.1 M NaClO4, pH adjusted to
3.1 with H3PO4; B: MeCN (0.1% TFA). Gradient: 0 min: 73/
27; 2 min: 73/27; 32 min: 70/30; 42 min: 50/50; 42.1 min:
73/27; 50 min: 73/27 min. Flow rate: 1.5 mL/min.
Product purity for compound 1 prepared from 22 was
evaluated by HPLC using the following conditions: Column:
Waters X-bridge shield RP18, 3.5 μm, 4.6 mm × 150 mm ID.
Column temperature: 30 °C. Detection: UV @ 220 nm.
Sample dissolving solvent: DMSO. Mobile phase: A: water
(0.1% TFA); B: MeCN (0.1% TFA). Gradient: 0 min: 80/20; 8
min: 80/20; 8.5 min: 75/25; 12 min: 68/32; 20 min: 35/65; 22
min: 35/65; 22.1 min: 80/20; 30 min: 80/20. Flow rate: 1.0
mL/min.
Reaction completion and product purity for compounds 6,
13, 14, 20, and reaction completion for compound 22 were
evaluated by HPLC using the following conditions: Column:
Zorbax SB-CN 3.5 μm, 3 mm × 75 mm. Column temperature:
45 °C. Detection: UV @ 210 nm. Mobile phase: A: water
(0.05% TFA); B: MeCN. Gradient: 0 min: 95/5; 3.7 min:
5/95; 4.3 min: 5/95; 4.4 min: 95/5. Flow rate: 1.2 mL/min.
Product purity for compound 22 was evaluated by HPLC
using the following conditions: Column: Waters X-bridge
shield RP18, 3.5 μm, 4.6 mm × 150 mm ID. Column
temperature: 30 °C. Detection: UV @ 220 nm. Sample
dissolving solvent: DMSO. Mobile phase: A: water (0.1%
TFA); B: MeCN (0.1% TFA). Gradient: 0 min: 80/20; 8 min:
80/20; 8.5 min: 75/25; 12 min: 68/32; 20 min: 35/65; 22 min
35/65; 22.1 min: 80/20; 30 min: 80/20. Flow rate 1.0 mL/min.
Synthesis of 1-(3-(4-Nitrophenyl)propanoyl)azetidin-
2-one (13). To a solution of 3-(4-nitrophenyl)propanoic acid
(9, 440 g, 2.25 mol), 2-azetidinone (11, 240 g, 3.37 mol), and
DIPEA (1.46 kg, 11.3 mol) in MeCN (4.4 L) was added a 50
EXPERIMENTAL SECTION
■
Reaction completion for compound 1 was evaluated by HPLC
using the following conditions: Column: XBridge BEH130 C18
3.5 μm, 4.6 mm × 150 mm. Column temperature: 30 °C.
Detection: UV @ 210 nm. Mobile phase: A: 60/40 water/
MeOH (0.1% formic acid); B: 60/40 MeCN/MeOH (0.085%
formic acid). Gradient: 0 min: 95/5; 3 min: 95/5; 20 min: 60/
G
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
wt % solution of 1-propanephosphonic acid anhydride (T3P) in
EtOAc (2.15 kg, 3.38 mol) over 1 h at 20−25 °C. After stirring
at 22 °C for 20 h, HPLC analysis of the reaction mixture
showed 0.9% of unreacted 9. The reaction was concentrated
under reduced pressure to about 2 L and the residue was
dissolved in i-PrOAc (6.6 L) at 30−35 °C. The organic layer
was washed with 10% aqueous citric acid (4 L). The aqueous
layer was back-extracted with i-PrOAc (2.7 L), and to the
combined organic extracts was added Darco G-60 (90 g). The
mixture was stirred at 25 °C for 4 h and filtered through Celite.
The Celite filter was washed with i-PrOAc (0.5 L), and the
filtrates were concentrated under reduced pressure to about 1
L. 2-Propanol (2.2 L) was added, and the distillation was
resumed to a final volume of about 1 L. 2-Propanol (2.2 L) was
added, and the mixture was cooled to 3 °C and held at this
temperature for 1 h. The solid was filtered and washed with 2-
propanol (0.55 L). The wet cake was transferred back to the
reactor and dissolved in EtOAc (3.24 L). Darco G-60 (90 g)
was added, and the mixture was stirred at 25 °C for 2 h. The
suspension was filtered through Celite, and the Celite pad was
washed with EtOAc (0.5 L). The filtrates were concentrated
under reduced pressure to about 1 L, and 2-propanol (1.4 L)
was added. The solution was concentrated to about 1 L, and
additional 2-propanol (1.4 L) was added. The suspension was
cooled to 3 °C and held at this temperature for 1 h. The solid
was filtered, washed with 2-propanol (0.5 L), and dried at 30−
35 °C under vacuum for 8 h to give 273 g (49%) of
intermediate 13. HPLC retention time: 2.68 min. HPLC purity:
3 °C, stirred for 1 h and filtered. The solid was washed with i-
PrOAc (80 mL) and dried under vacuum at 40 °C for 12 h to
give 120 g (80% yield) of 6. GC headspace analysis showed
residual THF (0.13 wt %) and i-PrOAc (0.27 wt %). HPLC
retention time: 2.18 min. HPLC purity: 99.3% (area %). Mp:
139−140 °C. 1H NMR (400 MHz, DMSO-d6) δ 2.72−2.90 (m,
4H) 3.00 (t, J = 5.27 Hz, 2H), 3.41 (t, 2H), 4.12 (s, 2H), 4.16
(s, 2H), 7.10−7.16 (m, 2H), 7.47−7.53 (m, 2H), 9.77 (s, 1H),
12.84 (br s, 1H). 13C NMR (100 MHz, DMSO-d6) δ 29.16,
35.98, 36.84, 37.89, 68.54, 70.89, 120.09, 128.95, 136.24,
136.84, 166.20, 168.10. MS (ES+): 335 (M + H)+.
Synthesis of 2-(2-Oxo-2-(4-(3-oxo-3-(2-oxoazetidin-1-
yl)propyl)phenylamino)ethoxy)acetate Pentafluorophe-
nol Ester (20). To a cold (3−5 °C) solution of acid 6 (80 g,
239 mmol) and pentafluorophenol (19, 49 g, 266 mmol) in
THF (1.1 L) was added a solution of N,N′-dicyclohexylcarbo-
diimide (52 g, 252 mmol) in THF (400 mL) over 10 min. After
15 min, the mixture was warmed to 22 °C and stirred for 18 h.
After HPLC analysis showed complete reaction, the suspension
was filtered to remove the dicyclohexylurea byproduct, and the
solid was washed with THF (110 mL). The filtrates were
concentrated to about 1.8 L, and acetone (1.58 L) was added.
The suspension was cooled to 10 °C and stirred for 1.5 h. The
remaining dicyclohexylurea was filtered off, and the solid was
washed with acetone (25 mL). The filtrates were concentrated
to about 1.8 L, and 2-propanol (2.75 L) was added. The slurry
was stirred at 20 °C for 16 h, and the solid was filtered, washed
with 2-propanol (110 mL), and dried under vacuum at 40 °C
for 16 h to give 95 g (83%) of pentafluorophenol ester 20 as a
white solid. HPLC retention time: 3.25 min. HPLC purity: 96%
1
98.0% (area %). Mp: 104−106 °C. H NMR (400 MHz,
DMSO-d6) δ 2.88−3.05 (m, 6H), 3.42 (t, J = 5.27 Hz, 2H),
7.44−7.53 (m, 2H), 8.07−8.16 (m, 2H). 13C NMR (100 MHz,
DMSO-d6) δ 29.37, 36.05, 36.89, 37.00, 123.83, 130.06, 146.38,
149.49, 166.20, 169.38. MS (ES+): 249 (M + H)+.
1
(area %). Mp: 94−99 °C. H NMR (400 MHz, CDCl3) δ
2.85−3.07 (m, 6H), 3.53 (t, J = 5.27 Hz, 2H), 4.26 (s, 2H),
4.62 (s, 2H), 7.10−7.22 (m, 2H), 7.40−7.53 (m, 2H), 8.47 (s,
1H). 13C NMR (100 MHz, CDCl3) δ 29.37, 35.83, 36.50,
37.98, 67.89, 69.03, 71.46, 71.83, 120.08, 120.16, 128.97,
129.08, 135.07, 136.87, 165.03, 166.13, 166.24, 170.10. MS
(ES+): 501 (M + H)+.
Synthesis of 2-(2-Oxo-2-(4-(3-oxo-3-(2-oxoazetidin-1-
yl)propyl)phenylamino)ethoxy)acetic Acid (6). A 30-L
hydrogenation reactor was charged with nitro compound 13
(142 g, 0.57 mol), 10% Pd/C (14 g), and THF (8.52 L). The
mixture was hydrogenated (10 psig) at 25 °C for 1 h. After
HPLC analysis showed complete consumption of the starting
material, the mixture was filtered through a Celite-precoated
0.5-μm cartridge filter, and the filter was washed with THF (3
L). To the filtrates containing aniline 1427 was added diglycolic
anhydride (15, 75 g, 0.65 mol). After 15 min at 22 °C, HPLC
analysis of the mixture showed complete consumption of 14,
and the reaction was concentrated to 2 L under reduced
pressure. The residue was seeded with a small amount of acid 6
crystals (1.06 g), and after 15 min, i-PrOAc (1.78 L) was added
over a 1-h period. The suspension was concentrated under
reduced pressure to 1.5 L, and the residue was cooled to 3 °C.
After 1 h, the solid was filtered, washed with cold i-PrOAc (0.5
L), and dried under vacuum at 40 °C to give 160 g (84%) of
acid 6. GC headspace analysis showed residual THF (0.60 wt
%) and i-PrOAc (0.62 wt %).
This material was reworked as follows to improve color and
reduce the amount of residual solvent. Acid 6 (150 g, 0.45 mol)
was dissolved in THF (5.25 L) at 30 °C. Activated carbon
Darco KBB (30 g) was added, and the mixture was stirred for 1
h at this temperature. The suspension was filtered through
Celite, and the Celite pad was washed with THF (0.5 L). The
filtrates were concentrated under reduced pressure to 1.5 L, and
to the residue was added i-PrOAc (3 L) over 30 min. The
mixture was held at 20 °C for 2 h, and it was then concentrated
under reduced pressure to 1.5 L. The suspension was cooled to
Synthesis of Peptide−Linker Conjugate 1 from 20. To
a solution of peptide 3 (100 g, 35.2 mmol) in DMF (1.5 L) at
−15 °C was added NMM (4.8 mL, 43.6 mmol) followed by
pentafluorophenol ester 20 (50 g, 100 mmol) in small portions
over 5 min. The mixture was stirred at −15 °C for 7 h at which
point HPLC analysis showed less than 1% of unreacted 3. The
mixture was filtered through a 0.45 μm in-line filter, and the
filtrates were added to a second reactor containing MeCN
(12.8 L) at 20 °C over 5 min. A white precipitate formed
immediately. The first reactor was washed with DMF (100
mL), and this wash was added to the second reactor through
the in-line filter. The resulting slurry was stirred at 20 °C for 1
h, and the solid was filtered, washed with MeCN (3 × 1 L), and
dried under vacuum at 20 °C for 6 h to give 94 g (86%) of 1 as
a white solid. GC headspace analysis showed residual DMF (4.5
wt %) and MeCN (0.95 wt %). With the goal of reducing
solvent content to meet specifications, the solid was passed
through a #20 hand-sieve and then slurried in MeCN (10 L) at
22 °C for 1 h. The solid was filtered, washed with MeCN (2 ×
1 L), and dried at 30 °C for 24 h and at 40 °C for an additional
24 h to give 91 g of 1 (97% recovery, 83% overall yield). GC
headspace showed residual DMF (0.05 wt %) and MeCN (0.19
wt %). HPLC purity: 96.5% (area %). HPLC retention time:
17.99 min. MS (ES+): 3156.6 Da (M + H)+.
Synthesis of 4-Nitrophenyl2-(2-oxo-2-((4-(3-oxo-3-(2-
oxoazetidinyl)propyl)phenyl)amino)ethoxy) Acetate 22.
H
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(5) (a) Pang, Y.; Liu, J.; Wu, J.; Li, G.; Wang, R.; Su, Y.; He, P.; Zhu,
X.; Yan, D.; Zhu, B. Bioconjugate Chem 2010, 21, 2093. (b) Voit, B. I.;
Lederer, A. Chem. Rev. 2009, 109, 5924. (c) Carlmark, A.; Hawker, C.;
Hult, A.; Malkoch, M. Chem. Soc. Rev. 2009, 38, 352. (d) Saha, A.;
Ramakrishnan, S. Macromolecules 2008, 41, 5658. (e) Gao, C.; Yan, D.
To a cold (0−5 °C) solution of p-nitrophenol (21, 0.23 g, 1.65
mmol) and acid 6 (0.5 g, 1.5 mmol) in THF (10 mL) was
added DCC (0.325 g, 1.6 mmol, 1.05 equiv). The reaction was
stirred at room temperature overnight followed by filtration to
remove dicyclohexylurea byproduct. The filtrates were
evaporated to dryness under reduced pressure, and to the
residue was added 2-propanol (8 mL). The slurry was stirred at
0−5 °C for 1 h, and the solid was filtered, washed with 2-
propanol (5 mL), and dried under high vacuum overnight to
provide 0.531 g (68%) of PNP ester 22 as a pale-yellow solid.
HPLC purity: 97.4 (area %). HPLC retention time: 17.75 min.
1H NMR (400 MHz, CDCl3) δ 3.02 (m, 6H), 3.58 (t, J = 5.30
Hz, 2H), 4.30 (s, 2H), 4.57 (s, 2H), 7.21 (d, J = 8.08 Hz, 2H),
7.36 (d, J = 8.96 Hz, 2H), 7.50 (d, J = 8.25 Hz, 2H), 8.31 (dd, J
= 8.81 Hz, J = 1.41 Hz, 2H), 8.59 (s, 1H). 13C NMR (100
MHz, CDCl3) δ 29.52, 35.97, 36.61, 38.16, 68.93, 71.95,
120.17, 122.29, 125.56, 129.24, 135.35, 136.92, 145.13, 154.48,
165.11, 166.46, 167.86, 170.21. HRMS: m/z: Calcd for (M+ +
H) C22H22N3O8 456.1329; found: 456.1422.
́
Prog. Polym. Sci. 2004, 29, 183. (f) Tomalia, D. A.; Frechet, J. M. J. J.
Polym. Sci., Part A: Polym. Chem. 2002, 40, 2719. (g) Bosman, A. W.;
Janssen, H. M.; Meijer, E. W. Chem. Rev. 1999, 99, 1665.
(6) (a) Doppalapudi, V. R.; Lai, J.-Y.; Liu, B.; Liu, D.; Desharnais, J.;
Bhat, A. S.; Fu, Y.; Oates, B. D.; Gang, C.; Bradshaw, C. W. WO/
2009/136352 A1 20091112, CAN 151:571334 (b) Bradshaw, C.;
Sakamuri, S.; Fu, Y.; Oates, B.; Desharnais, J.; Tumelty, D. WO/2008/
081418 A1 20080710, CAN 149:168435. (c) Bradshaw, C.; Bhat, A.;
Lai, J.-Y.; Doppalapudi, V.; Liu, D. WO/2008/056346 A2 20080515,
CAN 148:568835.
(7) (a) Annathur, G. V.; Balu, P.; Finn, R. F.; Huang, J.; Laurent, O.
A.; Levin, N. J.; Luksha, N. G.; Martin, J. P., Jr.; Moskowitz, H.;
Palanki, M. S. S.; Pozzo, M. J.; Waszak, G. A.; Xie, J. WO/2013/
093720 A2 20130627, CAN 159:159296. (b) Ishino, T.; Palanki, M. S.
S.; Violand, B. N.; Das, T. K.; Hodge, T. S.; Levin, N. J.; Parsons, E. K.
WO/2012/059873 A2 20120510, CAN 156:628793.
Synthesis of Peptide−Linker Conjugate 1 from 22. To
a solution of PNP ester 22 (0.3 g, 0.66 mmol, 3 equiv) in DMF
(5 mL) was added peptide 3 (0.624 g, 0.225 mmol) as a solid.
The reaction mixture was stirred at 0−5 °C for 8 h, at which
point HPLC analysis showed complete consumption of peptide
3. The mixture was added dropwise to MeCN (150 mL), and
the resulting slurry was stirred at 20 °C for 3 h. The solid was
filtered, washed with MeCN (15 mL), and dried under high
vacuum for 16 h to afford 0.612 g (86%) of 1 as a white
powder. HPLC purity: 96.8 (area %). HPLC retention time:
16.5 min.
(8) In this context, the term “reconstitution” implies full redissolution
of a solid in an appropriate solvent.
(9) After this work was complete, it was found that part of this batch
showed a very small degree of molecular order as shown by PXRD
analysis. As a result, the solubility properties changed dramatically, and
this material did not dissolve in MeCN/H2O 1:1 (v/v).
(10) Our group has developed an in-house method for the
preparation of peptide 3. See: Van Alsten, J. G. 2010 American
Institute of Chemical Engineers Annual Meeting, Conference
Proceedings, Salt Lake City, UT, U.S.A., November, 2010.
(11) Magano, J.; Maloney, M. T.; Marcq, O. J.; Nadkarni, D. V.;
Pozzo, M. J.; Teixeira, J. J., Jr. WO/2013/093705 A2, CAN
159:166192.
ASSOCIATED CONTENT
■
(12) Palanki, M. S. S.; Bhat, A.; Lappe, R. W.; Liu, B.; Oates, B.;
Rizzo, J.; Stankovic, N.; Bradshaw, C. Bioorg. Med. Chem. Lett. 2012,
22, 4249.
S
* Supporting Information
1
Copies of H and 13C NMR spectra for compounds 6, 13, 14,
20, and 22. This material is available free of charge via the
Internet at http://pubs.acs.org.
(13) (a) Feledziak, M.; Michaux, C.; Urbach, A.; Labar, G.; Muccioli,
G. G.; Lambert, D. M.; Marchand-Brynaert, J. J. Med. Chem. 2009, 52,
7054. (b) Urbach, A.; Dive, G.; Tinant, B.; Duval, V.; Marchand-
Brynaert, J. Eur. J. Med. Chem. 2009, 44, 2071. (c) Urbach, A.; Dive,
G.; Marchand-Brynaert, J. Eur. J. Org. Chem. 2009, 1757. (d) Staub, I.;
Sieber, S. A. J. Am. Chem. Soc. 2008, 130, 13400. (e) Cainelli, G.;
Galletti, P.; Garbisa, S.; Giacomini, D.; Sartor, L.; Quintavalla, A.
Bioorg. Med. Chem. 2003, 11, 5391. (f) Cainelli, G.; Giacomini, D.;
Gazzano, M.; Galletti, P.; Quintavalla, A. Tetrahedron Lett. 2003, 44,
6269.
(14) (a) Youcef, R. A.; Boucheron, C.; Guillarme, S.; Legoupy, S.;
Dubreuil, D.; Huet, F. Synthesis 2006, 633. (b) Firestone, R. A.; Barker,
P. L.; Pisano, J. M.; Ashe, B. M.; Dahlgren, M. E. Tetrahedron 1990, 46,
2255.
(15) KF on alumina has been employed as base in several types of
transformations. See: (a) Rohman, Md. R.; Myrboh, B. Tetrahedron
Lett. 2010, 51, 4772 and references therein. (b) Clark, J. H.; Farmer, T.
J.; Macquarrie, D. J. ChemSusChem 2009, 2, 1025.
AUTHOR INFORMATION
Corresponding Author
*E-mail: Javier.Magano@Pfizer.com
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Melissa Casteel and Chris Seadeek (Materials
Science), Peter Rose and Kevin Girard (Crystallization Group),
Rich Barnhart and David Bill (Process Safety Laboratory), Dan
D’Amico and David LaFleur (Supply Chain Analytical
Control), Dave Nelson (Analytical Research & Development),
April Xu, Lawrence Chen, and Julius Lagliva (Biotherapeutics
Analytical Research & Development), Joe Collins (Bioprocess
Research & Development) and Groton’s kilo laboratory staff
for their help during the implementation of this project.
(16) García, A. L. L. Synlett 2007, 1328.
(17) T3P is also available as a ∼50% solution in DMF from Sigma-
Aldrich.
(18) Several commercially available activated carbons were screened
(CGSP, SX-Plus, Darco S-51HF, E Supra USP, SX-Ultra, CASP,
Darco G-60) and Darco G-60 was chosen on the basis of cost and the
low amount of impurity 16 produced during the subsequent nitro
reduction step.
(19) The quality of acid 9 varied greatly depending on the source and
was critical to obtain reproducible results. When lower-purity 9 (85%)
was employed during a cGMP gram-lab campaign, two carbon
treatments were necessary to remove highly colored impurities, which
caused the yield of coupling product 13 to drop to 49%.
REFERENCES
■
(1) Hermanson, G. T. Bioconjugate Techniques, 2nd ed.; Academic
Press: San Diego, CA, 2008.
(2) Veronese, F. M.; Morpurgo, M. Farmaco 1999, 54, 497.
(3) Le Sann, C. Nat. Prod. Rep. 2006, 23, 357.
(4) (a) Hermentin, P.; Seiler, F. R. Behing Inst. Mitt. 1988, 197.
(b) Hermentin, P.; Doenges, R.; Gronski, P.; Bosslet, K.; Kraemer, H.
P.; Hoffmann, D.; Zilg, H.; Steinstraesser, A.; Schwarz, A.; Kuhlmann,
L.; Luben, G.; Seiler, F. R. Bioconjugate Chem. 1990, 1, 100.
̈
I
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(20) This large amount of solvent was needed to meet the minimum
stir volume requirement of a 30-L hydrogenator in our kilo laboratory
facility.
(21) (a) Almiento, G. M.; Balducci, D.; Bottoni, A.; Calvaresi, M.;
Porzi, G. Tetrahedron: Asymmetry 2007, 18, 2695. (b) Shin, I.; Jung,
H.-j.; Lee, M.-r. Tetrahedron Lett. 2001, 42, 1325. (c) Journet, M.; Cai,
D.; DiMichele, L. M.; Hughes, D. L.; Larsen, R. D.; Verhoeven, T. R.;
Reider, P. J. J. Org. Chem. 1999, 64, 2411. (d) Menger, F. M.; Rourk,
M. J. J. Org. Chem. 1997, 62, 9083. (e) Schneider, J. P.; Kelley, J. W. J.
Am. Chem. Soc. 1995, 117, 2533. (f) Boger, D. L.; Yohannes, D. J. Org.
Chem. 1990, 55, 6000.
(22) The source of TFA was the chromatographic purification of
peptide 3 using this acid as modifier.
(23) After the final conditions to run this step had been identified,
the purity of isolated product 1 was in all cases almost identical to the
purity of the peptide 3 starting material employed in the reaction. A
specification was set of ≥98% chemical purity for 3 so that the
chemical purity of the isolated 1 could be well above the desired ≥95%
specification.
(24) During the reviewing process for this manuscript, one of the
reviewers suggested that the formation of the PFP ester impurities
could be explained through the formation of a mixed anhydride
resulting from the reaction between PFP ester 20 and the carboxylic
acid functionalities on peptide 3. The resulting PFP coproduct would
then cleave the mixed anhydride, leading to the observed PFP
impurities. Presumably, PNP ester 22 would be less reactive than 20,
and mixed anhydride formation would not take place.
(25) Both peptide 3 and conjugate 1 tended to form thick gels in
aprotic, polar solvents such as DMF, DMAc, and DMSO after a few
hours, either in the presence or absence of water.
(26) In some experiments, only three pentafluorophenol ester
impurities could be detected by HPLC.
(27) Analytical data for 1-(3-(4-aminophenyl)propanoyl)azetidin-2-
one HCl (14·HCl): HPLC retention time: 3.25 min. Melting point:
>250 °C. 1H NMR (400 MHz, DMSO-d6) δ 2.80−2.92 (m, 4H), 3.00
(t, J = 5.27 Hz, 2H), 3.41 (t, J = 5.27 Hz, 2H), 7.16−7.35 (m, 4H),
10.24 (br s, 3H). 13C NMR (100 MHz, DMSO-d6) δ 29.10, 36.04,
36.87, 37.59, 123.67, 129.97, 130.21, 141.07, 166.24, 169.64. MS
(ES+): 219 (M + H)+.
J
dx.doi.org/10.1021/op4002998 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX