pubs.acs.

org/OPRD Article

Impurity Dimethylcarbamoyl-OBt Detected in an API Induced by the

Degradation of HBTU in DMF
Yi Yang,* Lena Hansen, Jörgen Kjellgren Sjögren, Ileana Rodríguez León, Jean-Marie Receveur,
and Fabrizio Badalassi*
Cite This: Org. Process Res. Dev. 2021, 25, 1923−1931 Read Online

ACCESS Metrics & More Article Recommendations *

sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Downloaded via FERRING PHARMACEUTICALS AS on August 23, 2021 at 07:17:47 (UTC).

ABSTRACT: A non-peptide-related impurity with an abundance of 0.09% was detected in a batch of Ferring Pharmaceuticals’
peptide active pharmaceutical ingredient (API) Peptide Q. A dedicated simulative reaction and comparison with the references
revealed that the impurity was induced by the degradation of the uronium coupling reagent HBTU in the dimethylformamide
(DMF) solution at the HBTU-directed peptide amidation cyclization step. The structure of the subject non-peptide-related impurity
has been confirmed as 1-[((dimethylamino)carbonyl]oxy)-1H-benzotriazole or its isomer 1-(dimethylcarbamoyl)-1H-1,2,3-
benzotriazol-3-ium-3-olate (both are abbreviated as dimethylcarbamoyl-OBt in this paper). The formation of dimethylcarbamoyl-
OBt could also be realized in N-methyl-2-pyrrolidone (NMP) or acetonitrile (ACN) as the organic solvent. This finding is relevant
to a myriad of syntheses of peptides and small molecules involved in uronium coupling reagent-mediated amidation/esterification
reactions, such as cyclization, side-chain modification, and segment condensation. In such cases, particularly those that proceed with
the subject HBTU-mediated reaction in DMF, NMP, or ACN as the very last step of the whole synthetic process, scrutiny of the API
for dimethylcarbamoyl-OBt as an impurity should be considered.
KEYWORDS: non-peptide-related impurity, peptide amidation cyclization, uronium coupling reagent degradation, HBTU, DMF,
1-[((dimethylamino)carbonyl]oxy)-1H-benzotriazole, 1-(dimethylcarbamoyl)-1H-1,2,3-benzotriazol-3-ium-3-olate,
dimethylcarbamoyl-OBt, nucleophilic aprotic solvent, Vilsmeier−Haack reaction

■ INTRODUCTION
Product quality always outstands as one of the most critical
Peptide Q, a cyclic peptide therapeutic of Ferring
Pharmaceuticals, undergoes amidation cyclization as the last
USP step, splicing the reciprocal amino and carboxylate group
aspects for therapeutics manufacturing, including peptide
from the linear Peptide Q precursor intramolecularly in the
active pharmaceutical ingredients (APIs). Impurities formed DMF solution by the coupling reagent HBTU. N-methyl
in the manufacturing process, particularly those that survived morpholine (NMM) is employed as the auxiliary base in this
the purification, are pivotal to product quality. In addition to reaction. This process is depicted in Figure 1.
systematic scrutiny for peptide-related impurities, our attention The derived Peptide Q-DMF cyclization solution was
on non-peptide-related impurities (small molecules) has diluted by an AcOH aqueous solution before being subjected
increased over the years due to the observation that these to chromatographic purification. A non-peptide-related im-
types of impurities can, more often than what can be purity with an abundance of 0.09% was detected in the final
anticipated, survive the purification steps in downstream
processing and ultimately end up in the final APIs.
Many cyclic peptide therapeutics undergo amidation
cyclization in the solution as the last manufacturing upstream
processing (USP) step. Uronium or phosphonium reagents
such as HBTU, TBTU, or PyBOP are frequently utilized as
coupling reagents to this end, and the subject cyclization
reactions generally proceed in dimethylformamide (DMF) as
the solvent. It is thus highly relevant to understand the fate of
Figure 1. Scheme of Peptide Q cyclization.
the coupling reagent under the subject reaction conditions,
elucidate the structures of the degradants, and assess their
toxicological effects, not least for the peptide therapeutics Received: May 11, 2021
GMP manufacturing. The findings will be highly instructive in Published: July 27, 2021
understanding the toxicological data, the establishment of the
API specification, the development of the analytical methods
for the product release, the prediction of the API stability, and
the guidance for the formulation.

© 2021 American Chemical Society https://doi.org/10.1021/acs.oprd.1c00168

1923 Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 2. UV chromatogram (upper) and TIC (total ion chromatogram, bottom) of the HBTU/DIPEA/DMF solution. Peak 2: HBTU; peak 3:
HOBt; peak 4: TMU; and peak 6: dimethylcarbamoyl-OBt.

API of Peptide Q after the chromatographic purification, which Peptide Q, the subject non-peptide-related impurity was
was very close to the upper specification limit (USL) of 0.1%. entirely purged by the corresponding chromatographic
This impurity remained unknown at the time point of purifications from Peptide Y. The formation of the subject
detection. It seemed that the subject non-peptide-related impurity through the degradation of HBTU in DMF in the
impurity had a very similar chromatographic property to that presence of either NMM or DIPEA, identified in both Peptide
of Peptide Q and was detected in many chromatographic Q and Peptide Y projects, could be extrapolated and
fractions containing the target product Peptide Q. Attempts to generalized to the API manufacturing involving amide bond
elucidate the structure of the subject impurity through liquid formation directed by the uronium coupling reagent and the
chromatography/mass spectrometry (LC/MS) and LC/ auxiliary tertiary base in DMF.
nuclear magnetic resonance (NMR) did not lead to an
unambiguously definitive conclusion at that stage.
Non-peptide-related impurities in peptide APIs are generally
■ RESULTS
Simulation of HBTU Degradation in DMF. Investigation
treated at Ferring Pharmaceuticals in a consistent QbD manner on the fate of HBTU in the process of Peptide Q cyclization
through the prediction of impurity formation, the impurity fate was designed and conducted with an effort to understand the
investigation by simulation, and the verification. Within the formation of non-peptide-related impurities at this step,
scope of the detected non-peptide-related impurity in Peptide particularly for the one detected in the Peptide Q API with
Q, it was assumed that it was highly probably formed at the an alerting abundance of 0.09%. In the prototypical process,
very last USP step, i.e., amidation cyclization in DMF, since it linear Peptide Q was dissolved in DMF. Then, the derived
was not detected in the linear Peptide Q starting material. linear peptide solution was added slowly to the HBTU/
Therefore, the simulation reaction was designed and NMM/DMF solution, where amidation cyclization of the
conducted to mimic the reaction of Peptide Q cyclization linear Peptide Q took place.
but in the absence of linear Peptide Q to disentangle its It is to note that simulation of HBTU degradation was
potential interferences on the subject non-peptide-related conducted in the context of another affected peptide project,
impurity formation and its analyses, as well as to enrich the Peptide Y. HBTU/DMF solution, neat DIPEA, and neat DMF
impurity content. were separately loaded in three syringes, and they were
In this study, the designed degradation of HBTU in DMF synchronically added through a syringe pump to a cyclization
under the subject conditions gave rise to the formation of a reactor encompassing DMF solvent within 120 min. The
degradant that matched the subject non-peptide-related peptide solute was eliminated from the corresponding DMF
impurity detected in the Peptide Q API. The structure of solution in the third syringe to dodge the potential interference
the impurity was elucidated based on the comparison of the of the peptide and isolate the effect of the HBTU degradation.
LC/MS and MS/MS data of the impurity with those from the Otherwise, all of the other reagents and parameters were
reference materials. The successful identification of the identical to those of the Peptide Y cyclization process. The
impurity will serve as a starting point for a toxicological reaction mixture was then acidified by a 1% AcOH aqueous
assessment in Peptide Q drug development. Moreover, this solution and stored overnight before LC/MS analysis.
impurity was also detected in the crude product of another The decomposition of HBTU in DMF in the presence of
Ferring Pharmaceuticals’ cyclic peptide in development DIPEA was monitored by LC/MS (Figure 2). HBTU
(Peptide Y), which also underwent amidation cyclization by underwent intensive degradation in the simulation reaction,
HBTU/N,N-diisopropylethylamine (DIPEA) in DMF. Unlike and only 39% of HBTU (peak 2) survived the decomposition.
1924 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

The LC/MS detected predominant degradants HOBt (40.2%, to a reliably affirmative conclusion on its origin, despite the fact
peak 3) and TMU (tetramethylurea, 13.7%, peak 4) along with that the absolute isomer could not be determined at the
a peculiar degradant with a distinctive molecular weight of current stage (Supporting Information).
206.0 Da (peak 6). The dimethylcarbamoyl-OBt impurity, together with HOBt
1-[((Dimethylamino)carbonyl]oxy)-1H-benzotriazole (di- and TMU, was also detected in four different batches of
methylcarbamoyl-OBt, Figure 3) was first tentatively assigned Peptide Y crude products after the amidation cyclization in
HBTU/DIPEA/DMF. However, the dimethylcarbamoyl-OBt
impurity in the Peptide Y crude product was nonetheless
entirely removed by the corresponding chromatographic
purifications.
Correlation of Dimethylcarbamoyl-OBt Formation
with Solvent, Basicity, and Coupling Reagent. The
correlation between the dimethylcarbamoyl-OBt formation
and the reaction conditions has been investigated in this study.
Figure 3. Chemical structure of 1-[((dimethylamino)carbonyl]oxy)- Variations on the solvent, base quantity, and the coupling
1H-benzotriazole (dimethylcarbamoyl-OBt adduct). reagent have been introduced, and their impacts on
dimethylcarbamoyl-OBt formation were explored. HBTU or
PyBOP were opted as the coupling reagent, whereas DMF, N-
as the degradant given its molecular weight and the structure methyl-2-pyrrolidone (NMP), dichloromethane (DCM),
relevance. The HBTU degradation experiment was conducted toluene, H2O, acetonitrile (ACN), trifluoroacetic acid
at an elevated temperature of 50 °C under otherwise identical (TFA)/H2O/ACN, H2O/ACN, H2O/MeOH, and H2O/
conditions to verify the proposed chemical structure by EtOH were tested as solvents. The selected coupling reagent
accelerating the reaction. The subject degradant was put was stressed in a specific solvent in the presence or absence of
under comparison with the commercially available reference DIPEA under the ambient temperature or an elevated
material 1-[((dimethylamino)carbonyl]oxy)-1H-benzotriazole. temperature of 50 °C for a period of 3, 5, or 120 h. The
The comparison of the HBTU degradant and the reference derived solution was subjected to LC/MS analysis and
material 1-[((dimethylamino)carbonyl]oxy)-1H-benzotriazole compared with that of the dimethylcarbamoyl-OBt references.
(dimethylcarbamoyl-OBt) by LC/MS analysis indicates The results of the degradation study are summarized in Table
strikingly high similarity. The retention time and the MS 1. It is to note that HBTU (or PyPOB), HOBt, and
fragmentation pattern strongly corroborate that HBTU was dimethylcarbamoyl-OBt were normalized by the UV chroma-
partially degraded into dimethylcarbamoyl-OBt in the DIEPA/ togram from the LC/MS quantification.
DMF solution. Strong solvent dependence of dimethylcarbamoyl-OBt
Based on the information provided by the simulated HBTU formation was explicitly obtained through the designed
degradation in DMF, a thorough comparison of 1- degradation experiments. No dimethylcarbamoyl-OBt degra-
[((dimethylamino)carbonyl]oxy)-1H-benzotriazole (CAS No. dant could be detected in the solvent other than DMF, NMP,
39968-13-3) and its isomer 1-(dimethylcarbamoyl)-1H-1,2,3- and ACN (trace was formed in a H2O/ACN solution from
benzotriazol-3-ium-3-olate (CAS No. 305862-39-9) with the exp. 13). It is somewhat surprising to detect dimethylcarba-
detected non-peptide-related impurity in Peptide Q was moyl-OBt in exp. 2 and 15, where NMP and ACN functioned
performed. The two isomers are arguably interconvertible as solvents, respectively, since there is no intrinsic
(Figure 4). Under the five specific reversed-phase high- dimethylcarbamoyl moiety in these two solvents. The
performance liquid chromatography (RP-HPLC) conditions, formation of dimethylcarbamoyl-OBt in NMP and ACN
the subject non-peptide-related impurity from Peptide Q might be attributed to an essential characteristic shared by
unanimously coeluted with the two dimethylcarbamoyl-OBt DMF, NMP, and ACN. The putative underlying mechanism
isomers, unequivocally corroborating its identity (one RP- thereof will be delineated in the Discussion session.
HPLC result is displayed in Figure 5). On top of the The base is imperative for the subject degradation, and the
chromatographic investigation, NMR spectra of the subject formation of dimethylcarbamoyl-OBt was almost entirely
non-peptide-related impurity and the reference 1- suppressed in the void of DIPEA under otherwise identical
[((dimethylamino)carbonyl]oxy)-1H-benzotriazole could lead conditions (exp. 5 vs exp. 1). The stability of HBTU in DMF

Figure 4. Chemical structure of dimethylcarbamoyl-OBt isomers.

1925 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

Figure 5. Overlay of RP-HPLC traces of reference 1-[(dimethylamino)carbonyl]oxy-1H-1,2,3-benzotriazole, 1-(dimethylcarbamoyl)-1H-1,2,3-

benzotriazol-3-ium-3-olate, and Peptide Q.

Table 1. HBTU Degradation under Various Conditions

HBTU
exp. base temp. time or dimethylcarbamoyl-
no. substrate base equiv solvent (°C) (h) PyBOP HOBt OBt adduct other degradant
1 HBTU DIPEA 4.6 DMF 50 3 0% 98.4% 1.6%
2 HBTU DIPEA 4.6 NMP 50 3 0% 67.7% 0.3% unknown compound (32.0%)
3 HBTU DIPEA 4.6 DCM 50 8 0% 100% 0%
4 HBTU DIPEA 4.6 toluene 50 3 0% 100% 0%
5 HBTU no base 0 DMF 50 3 99.3% 0.7% 0%
6 HOBt DIPEA 4.6 DMF 50 3 N.A 100% 0%
7 PyBOP DIPEA 4.6 DMF 50 3 0% 96.2% 0% tris(pyrrolidinophosphine) oxide
(3.8%)
8 HBTU no base 0 TFA/ACN/H2O (0.067/33/67) RT 120 91.9% 8.1% 0%
9 HBTU DIPEA 35 DMF 50 3 0% 98.0% 2.0%
10 HBTU DIPEA 77 DMF 50 3 0% 98.8% 1.2%
11 HBTU DIPEA 4.6 H2 O 50 5 0% 100% 0%
12 HBTU DIPEA 4.6 EtOH/H2O (2:1) 50 5 0% 100% 0%
13 HBTU DIPEA 4.6 ACN/H2O (1:1) 50 3 0% 99.98% 0.02%
14 HBTU DIPEA 4.6 MeOH/H2O (1:1) 50 3 0% 100% 0%
15 HBTU DIPEA 4.6 ACN 50 3 0% 71.7% 0.73% m/z 175.05, probable 2-
(benzotriazole-1-yl-oxy)-
acetonitrile (27.6%)
16 HOBt DIPEA 4.6 DMF 50 3 0% 100% 0%
and
TMU

in the absence of base is also surprisingly high even at the incapable of accommodating its formation in DMF in the
elevated temperature, as indicated by exp. 5 that the majority presence of DIPEA (exp. 6). A phosphonium-type coupling
of HBTU (99.3%) remained intact at 50 °C after 3 h. HBTU reagent PyBOP also underwent thorough degradation in DMF
underwent swift degradation in an alkaline aqueous solution in the presence of DIPEA after 3 h at 50 °C (exp. 7). However,
(exp. 11−14), but no dimethylcarbamoyl-OBt degradant could the dimethylcarbamoyl-OBt adduct was not formed as a
be formed. The predominant degradation pattern in these consequence of PyBOP degradation in DMF. Tris-
experiments was the hydrolysis of HBTU to HOBt and TMU. (pyrrolidinophosphine) oxide was detected in the PyBOP
HBTU in an acidic aqueous solution (exp. 8), on the contrary, degradation solution instead.
remained distinctively stable, and the majority of HBTU Reactivity of Dimethylcarbamoyl-OBt. The reactivity of
(91.9%) survived the degradation in a solution mixture TFA/
dimethylcarbamoyl-OBt toward nucleophiles has also been
ACN/H2O (v/v/v, 0.067/33/67) that simulates an eluent of
investigated in this study. Potential hydrolysis of dimethylcar-
RP-HPLC. An increase of the DIPEA quantity from 4.6 equiv
(exp. 1) to 35 equiv (exp. 9) and 77 equiv (exp. 10) did not bamoyl-OBt and its susceptibility toward nucleophilic addition
boost the formation of dimethylcarbamoyl-OBt. This phenom- by the amino group were tested. In these experiments,
enon could be attributed to the inherent lability of dimethylcarbamoyl-OBt was dissolved in various aqueous
dimethylcarbamoyl-OBt and its degradation in the basic solution mixtures in the presence of DIPEA. Nucleophiles
milieu. A dedicated investigation in this study on the lability such as phenylalanine, glycine, tyrosine, or aniline were added
of dimethylcarbamoyl-OBt also sustains such a claim. to the reaction mixture. Transformation of dimethylcarbamoyl-
The electrophilicity of HBTU is paramount for the OBt at the ambient temperature was recorded, and the results
formation of dimethylcarbamoyl-OBt in DMF since HOBt is are summarized in Table 2.
1926 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

It is revealed by the data that dimethylcarbamoyl-OBt

days

20.1
5
underwent hydrolysis in the aqueous solution in the presence
of DIPEA. The kinetics of the subject hydrolysis seems to be
24 h

5.6
2.8
significantly dependent on the organic solvent constituent, as it
was much faster in MeOH/H2O (v/v, 1:1, exp. 1-2) than in
23 h

3.0
DMF/H2O (exp. 1-1) or iPrOH/H2O (exp. 1-3) under
dimethylcarbamoylamine product

otherwise identical conditions. The degradants of dimethyl-

19 h

99.1
carbamoyl-OBt hydrolysis include HOBt. The reciprocal
degradant should be dimethyl carbamic acid, which was

Dimethylcarbamoyl glycine is not integrated in RP-HPLC due to the elution at the dead time of the applied RP-HPLC. bHOBt degradant is not integrated in RP-HPLC.
further decomposed to dimethylamine and CO2.1 Degradation
5h

0.7
1.2
0.6
of dimethylcarbamoyl-OBt was significantly retarded in the
absence of base, as indicated by exp. 2 and 3-1 that the

14.0
4h

integrity of dimethylcarbamoyl-OBt was largely preserved

45.4 when DIPEA was excluded from the reaction mixture, and
10.2
2h

glycine or weak base aniline was supplemented as a

nucleophile. This conclusion is evidently sustained by exp. 6,
0.2
0.3
0.1

29.2

6.6
1h

in which the decomposition of dimethylcarbamoyl-OBt was

thoroughly suppressed in a mixture solution consisting of
0.5 h

TFA/ACN/H2O (v/v/v, 0.067/33/67), simulating the

0.1
0.2
0.1

17.8

chromatographic purification environment. Dimethylcarbamo-

yl-OBt was fully stable in such an acidic aqueous solution even
days

0.1
5

after 24-h stressing. This attribute also explains how

dimethylcarbamoyl-OBt formed at Peptide Q cyclization
75.2
87.2
24 h

100

could survive the subsequent chromatographic purification

and end up in the purified Peptide Q. It is to note that
tetramethylurea has been consistently detected in dimethyl-
23 h

86.1

carbamoyl-OBt hydrolysis. The mechanism thereof will be

elucidated in the Discussion section.
21 h
37.6

96.3
remaining dimethylcarbamoyl-OBt (%)

Despite the restrained reactivity of dimethylcarbamoyl-OBt

in the absence of auxiliary tertiary base and the poor
20 h

0.1
31.7

nucleophilicity of aniline, formation of dimethylcarbamoyl

aniline indeed proceeded slowly in exp. 2. On the contrary,
19 h

when DIPEA was added to exp. 3-2, dimethylcarbamoyl-OBt

0.9

was rapidly consumed by glycine even though the assumed

product dimethylcarbamoyl glycine could not be reliably
57.7

96.4
94.5
97.1
5h

100

quantified due to its elution at the dead time of the applied RP-
HPLC method. When glycine was substituted by phenyl-
12.4
55.0

98.8

40.6
4h

alanine in exp. 4, the derived product dimethylcarbamoyl

phenylalanine was unambiguously detected and quantified.
Dimethylcarbamoyl-OBt in exp. 4 was almost quantitatively
21.4
64.4

54.6

55.6
2h

consumed (>99%) at the ambient temperature after 19 h.

Similar reactivity toward tyrosine was also observed in exp. 5.
71.1

99.1
98.6
99.3
99.5
1.3

70.8

69.9

■
1h

100

DISCUSSION
71.8

99.4
99.2
99.6
99.7

82.2
0.5 h

100

The predominant degradation of HBTU is the base-catalyzed

Table 2. Reactions of Dimethylcarbamoyl-OBt

hydrolysis, splitting HBTU into HOBt and tetramethylurea.

Another major degradant, i.e., 1-[((dimethylamino)carbonyl]-
MeOH/H2O (1:1)

MeOH/H2O (1:1)
iPrOH/H2O (1:1)

iPrOH/H2O (1:1)
iPrOH/H2O (1:1)

iPrOH/H2O (1:1)

iPrOH/H2O (1:1)

oxy)-1H-benzotriazole or its isomer 1-(dimethylcarbamoyl)-

DMF/H2O (1:1)

DMF/H2O (1:1)

TFA/ACN/H2O
(0.067/33/67)
solvent

1H-1,2,3-benzotriazol-3-ium-3-olate (both denoted as dime-

thylcarbamoyl-OBt), was detected by LC/MS. The scheme
and putative mechanism of the observed HBTU degradation in
DMF in the presence of base are proposed in Scheme 1.
The formation of dimethylcarbamoyl-OBt is presumably
initiated by DMF as a nucleophile. The first step of the subject
DIPEA

DIPEA

DIPEA

DIPEA
12 equiv

12 equiv

12 equiv
base
6 equiv

degradation process might be similar to that of the Vilsmeier−

Haack reaction. DMF directs a nucleophilic attack at the highly
no

no

no

reactive electrophile HBTU in the presence of DIPEA and

reactant

gives rise to the formation of Vilsmeier salt analogue 1. Such a

aniline
8 equiv

8 equiv

8 equiv

8 equiv
Phe

Tyr
Gly

Vilsmeier salt analogue has been detected in HBTU

degradation in the DIPEA/NMP solution (reaction 2 from
Table 1) by LC/MS analysis (m/z 333.20, Scheme 2),
3-1a
3-2a
exp.
no.
1-1
1-2
1-3
2-1
2-2
2-3

4b

underpinning the plausibility of the postulated mechanism.

a

1927 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

Scheme 1. Proposed Mechanism of HBTU Degradation in DIPEA/DMF

Scheme 2. Formation of an Intermediate m/z 333.20 in underlying driving force for the formation of dimethylcarba-
HBTU Degradation in the DIPEA/NMP Solution moyl-OBt in NMP and ACN could also be attributed to their
intrinsic characteristic as an aprotic solvent with weak
nucleophilicity.6 The mechanism is postulated in Scheme 3.
Analogous to the DMF-involved reaction depicted in
Scheme 1, the first step of HBTU degradation involves the
nucleophilic attack of ACN or NMP on HBTU to form the
corresponding nitrilium 3 and imidate 5, respectively. The
nucleophilic attack of ACN on cationic HBTU is similar to the
first step of the Ritter reaction7 (the carbon atom from the
−C(N(CH3)2)2 moiety is partially positively charged through
the resonance and the isomerization between the uronium and
aminium species8). It is to note that compound 5 and its
The formed Vilsmeier salt analogue 1 could accommodate a isomers (m/z 333.20) have been detected by the LC/MS
nucleophilic attack from the released HOBt anion and form an analysis of the corresponding HBTU/DIPEA/NMP mixture.
unstable intermediate 2,2 which was hydrolyzed to dimethyl- Intermediates nitrilium 3 and imidate 5 could be stabilized by
carbamoyl-OBt under basic conditions. the resonance through delocalizing the positive charge between
Genotoxic impurity dimethylcarbamoyl chloride (DMCC)3
the subject nitrogen atom and the adjacent carbon atom.
as a minor byproduct from acid chloride preparation has been
Unlike the mechanism proposed in Scheme 1, intermediates 3
verified to follow a similar mechanism through the undesired
and 5 not only experience the OBt leaving (which might be the
function between DMF and a chlorination reagent such as
thionyl chloride.4 Unintended reactions between DMF as a predominant pathway) but also 1,3-amino group migration9 to
solvent and the coupling reagents are not uncommon in give isomers 4 and 6, respectively, that are stabilized by the
peptide chemistry, normally leading to formylation side corresponding resonances. Hydrolysis of 4 or 6 gives rise to the
reactions as consequences.5 formation of dimethylcarbamoyl-OBt. Alternatively, the
It is noted that dimethylcarbamoyl-OBt was also formed in spontaneous collapse of 6 may also form dimethylcarbamoyl-
NMP and ACN (exp. 2 and 15 in Table 1), although these two OBt and the reciprocal methylenediamine cation. In principle,
solvents are inherently void of the dimethylcarbamoyl moiety dimethylcarbamoyl-OBt formation in DMF could also follow
in their chemical structures. The formation of dimethylcarba- such a pathway.
moyl-OBt in NMP and ACN should not be attributed to the The common underlying characteristic of DMF, ACN, and
function between HOBt and TMU. Stressing of HOBt and NMP accountable for the dimethylcarbamoyl-OBt formation is
TMU in DMF in the presence of DIPEA at 50 °C for 3 h could their weak nucleophilicity as an aprotic solvent. The
not generate any detectable dimethylcarbamoyl-OBt in this nucleophilicity of DMF, ACN, and NMP facilitates their
study (exp. 16 in Table 1). nucleophilic addition to HBTU as the initiation step of HBTU
Apparently, the formation of dimethylcarbamoyl-OBt from degradation. Even though protic solvents like water, ethanol,
HBTU degradation in NMP and ACN could not proceed and methanol could also initiate an analogous nucleophilic
through the mechanism proposed in Scheme 1, which should attack on HBTU, the derived adducts almost exclusively
be unique to DMF. Despite such structural disparity, the collapse to HOBt and the reciprocal urea or O-alkyl isourea.
1928 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

Scheme 3. Putative Mechanism of ACN- and NMP-Induced HBTU Degradation and Dimethylcarbamoyl-OBt Formation

Scheme 4. Scheme of Dimethylcarbamoyl-OBt Reactions

On the contrary, the solvent−HBTU adduct developed from and the derived isomeric cation is hydrolyzed or spontaneously
an aprotic nucleophilic solvent such as DMF, NMP, or ACN is collapsed to dimethylcarbamoyl-OBt. Considering the obser-
positively charged and stabilized by resonances. Moreover, it vations from the HBTU degradation in DMF, ACN, and NMP
can accommodate 1,3-migration of the dimethylamino group, and the underlying mechanism, it might be potentially
1929 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

plausible to generalize the dimethylcarbamoyl-OBt formation MS, NMR, and the comparison with the references confirmed
to the other aprotic nucleophilic organic solvents. Indeed, its identity as 1-[((dimethylamino)carbonyl]oxy)-1H-benzo-
dimethylcarbamoyl-OBt was also detected in the analogous triazole or its arguably interconvertible isomer 1-(dimethylcar-
degradation test of HBTU in dimethyl sulfoxide (DMSO) and bamoyl)-1H-1,2,3-benzotriazol-3-ium-3-olate.
acetone in this study (data not shown). The subject HBTU degradation might be accounted to its
The generated dimethylcarbamoyl-OBt impurity retains strong electrophilicity and the latent nucleophilicity from
appreciable electrophilicity toward nucleophiles such as the DMF. A process similar to the Vilsmeier−Haack reaction
amino group in the basic milieu. The amino group is could be induced between DMF and HBTU, leading to the
derivatized by dimethylcarbamoylation as a consequence, and formation of the Vilsmeier salt analogue, being further
dimethylcarbamoyl-OBt is degraded to HOBt. Dimethylcarba- transformed into dimethylcarbamoyl-OBt. As the last USP
moyl-OBt also undergoes hydrolysis under basic conditions. step of Peptide Q production, dimethylcarbamoyl-OBt
TMU and HOBt are formed in this process. The formation of generated in the Peptide Q cyclization reaction could not be
TMU could be attributed to the intermediate product dimethyl sufficiently purged by the subsequent chromatographic
carbamic acid decomposed to dimethylamine through purifications and thus contaminated purified Peptide Q as a
decarboxylation. The derivatization of dimethylcarbamoyl- non-peptide-related impurity. The formation of dimethylcar-
OBt by dimethylamine results in the formation of TMU. bamoyl-OBt is highly dependent on the solvent, pH, and the
The reactions of dimethylcarbamoyl-OBt with amine and type of coupling reagents. DMF accommodates its formation.
water are depicted in Scheme 4. Besides, ACN, NMP, acetone, and DMSO could also incubate
The formation of dimethylcarbamoyl-OBt has been verified dimethylcarbamoyl-OBt generation, although they are void of
to be solvent-dependent. Therefore, organic solvents that the dimethylcarbamoyl moiety in their chemical structures.
could not accommodate its formation might replace DMF to The base is paramount to the dimethylcarbamoyl-OBt
suppress the dimethylcarbamoyl-OBt generation if the target formation, which is thoroughly suppressed under neutral and
coupling reactions, such as amidation and esterification, could acidic conditions. A phosphonium-type coupling reagent
proceed equally in alternative organic solvents with com- PyBOP is incapable of giving rise to dimethylcarbamoyl-OBt
parable reaction kinetics and racemization on the subject under comparable conditions. Moreover, it is recommendable
carboxylic acid C2 chiral center, if any. Otherwise, an to quench the further dimethylcarbamoyl-OBt development
alternative type of coupling reagent could be considered to from the subject HBTU-mediated amidation reaction upon its
prevent dimethylcarbamoyl-OBt impurity formation. completion through acidification.
Last but not least, it needs to be underscored that It needs to be emphasized that dimethylcarbamoyl-OBt
dimethylcarbamoyl-OBt formation is not exclusively address- formation is not exclusively restricted to peptide cyclization.
ing peptide cyclization, albeit being detected at a Peptide Q Any reactions waging HBTU, or arguably comparable uronium
and Peptide Y cyclization step. A prerequisite is the analogue, as a coupling reagent in DMF, NMP, or ACN could
involvement of HBTU and an aprotic nucleophilic solvent potentially generate dimethylcarbamoyl-OBt adduct as a
like DMF in a condensation reaction, come what may, in byproduct. Therefore, their existence in the final API should
peptide cyclization, segment condensation, side-chain mod- be scrutinized with a suitable analytical method.
ification, or small molecule construction within the scope of Prediction of the mutagenicity of the dimethylcarbamoyl-
amidation, esterification, or other relevant reactions. Therefore, OBt adduct has been conducted by QSAR, and the compound
utmost attention is needed, particularly when the subject has been categorized as Class 5 of ICH M7. The Ames test is
reaction is the very last step of the synthesis. conducted subsequently to assess its mutagenicity accurately.
Prediction of the mutagenicity of the dimethylcarbamoyl-
OBt adduct has been conducted by the quantitative structure−
activity relationship (QSAR), and the compound has been
■ ASSOCIATED CONTENT
* Supporting Information
sı
categorized as Class 5 of ICH M7 (Supporting Information). The Supporting Information is available free of charge at
Specification of such impurity in the final API and drug https://pubs.acs.org/doi/10.1021/acs.oprd.1c00168.
product should be rationally established based on its toxicity
and the drug dosage. Also, the reactivity property of Experimental section; LC/MS analysis of the dimethyl-
dimethylcarbamoyl-OBt should be obtained to understand its carbamoyl-OBt impurity; NMR analysis of the isolated
potency to modify the parent molecule. Moreover, it is relevant dimethylcarbamoyl-OBt impurity and reference 1-
to test whether other uronium coupling reagents such as [((dimethylamino)carbonyl]oxy)-1H-benzotriazole; for-
TBTU and oxyma-based coupling reagents such as (1-cyano-2- mation of m/z 175.05 and 217.11 species in HBTU
ethoxy-2-oxoethylidenaminooxy)dimethylamino-morpholino- degradation in ACN and NMP; and QSAR evaluation
carbenium hexafluorophosphate (COMU) could follow a for dimethylcarbamoyl-OBt (PDF)
comparable decomposition pathway and generate analogous
impurities. ■ AUTHOR INFORMATION

■ CONCLUSIONS
A non-peptide-related impurity in Peptide Q API was detected
Corresponding Authors
Yi Yang − Chemical Development, Global Pharmaceutical
R&D, Ferring Pharmaceuticals A/S, DK-2300 Copenhagen
with an abundance of 0.09%, generated at the step of HBTU- S, Denmark; orcid.org/0000-0002-0003-8426;
mediated Peptide Q cyclization in the DMF solution. HBTU Phone: +45 28 78 73 74; Email: [email protected]
underwent significant degradation in the DIPEA/DMF Fabrizio Badalassi − Chemical Development, Global
solution in the simulation reaction. Besides the known Pharmaceutical R&D, Ferring Pharmaceuticals A/S, DK-
degradants HOBt and TMU, another then unidentified 2300 Copenhagen S, Denmark; orcid.org/0000-0003-
derivative was also detected in the reaction. LC/MS, MS/ 3250-6826; Email: [email protected]
1930 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931
Organic Process Research & Development pubs.acs.org/OPRD Article

Authors
Lena Hansen − Chemical Development, Global
Pharmaceutical R&D, Ferring Pharmaceuticals A/S, DK-
2300 Copenhagen S, Denmark
Jörgen Kjellgren Sjögren − Analytical Development
Methods & Characterization, Global Pharmaceutical R&D,
Ferring Pharmaceuticals A/S, DK-2300 Copenhagen S,
Denmark
Ileana Rodríguez León − Analytical DevelopmentMethods
& Characterization, Global Pharmaceutical R&D, Ferring
Pharmaceuticals A/S, DK-2300 Copenhagen S, Denmark
Jean-Marie Receveur − Chemical Development, Global
Pharmaceutical R&D, Ferring Pharmaceuticals A/S, DK-
2300 Copenhagen S, Denmark
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.oprd.1c00168

Notes
The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS
The authors thank Dr. F. J. Petersen from Eurofins
Spinnovation Analytical for the NMR analyses, and Svava
Osk Jónsdóttir, Ph.D. from Saxocon A/S for the prediction of
the mutagenicity of dimethylcarbamoyl-OBt by QSAR. The
authors also thank Anja Goldenbaek from Non-Clinical R&D
Department of Ferring Pharmaceuticals A/S to coordinate the
QSAR study.

■ REFERENCES
(1) Caplow, M. Kinetics of Carbamate Formation and Breakdown. J.
Am. Chem. Soc. 1968, 90, 6795−6803.
(2) De Luca, L.; Giacomelli, G.; Porcheddu, A. Beckmann
Rearrangement of Oximes under Very Mild Conditions. J. Org.
Chem. 2002, 67, 6272−6274.
(3) ICH M7(R1) Assessment and Control of DNA Reactive
(Mutagenic) Impurities in Pharmaceuticals To Limit Potential
Carcinogenic Risk.
(4) Levin, D. Potential Toxicological Concerns Associated with
Carboxylic Acid Chlorination and Other Reactions. Org. Process Res.
Dev. 1997, 1, 182.
(5) Yang, Y. Side Reactions on Amino Groups in Peptide Synthesis.
In Side Reactions in Peptide Synthesis; Academic Press, 2015; pp 177−
179.
(6) Hoff, B. H. Acetonitrile as a Building Block and Reactant.
Synthesis 2018, 50, 2824−2852.
(7) Jiang, D.; He, T.; Ma, L.; Wang, Z. Recent developments in
Ritter reaction. RSC Adv. 2014, 4, 64936−64946.
(8) Abdelmoty, I.; Albericio, F.; Carpino, L.; Foxman, B.; Kates, S.
Structural studies of reagents for peptide bond formation: Crystal and
molecular structures of HBTU and HATU. Lett. Pept. Sci. 1994, 1,
57−67.
(9) Chauhan, D. P.; Varma, S. J.; Vijeta, A.; Banerjee, P.; Talukdar,
P. A 1,3-amino group migration route to form acrylamidines. Chem.
Commun. 2014, 50, 323−325.

1931 https://doi.org/10.1021/acs.oprd.1c00168
Org. Process Res. Dev. 2021, 25, 1923−1931