Deamidation in Moxetumomab Pasudotox Leading to Conformational Change And
bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Deamidation in Moxetumomab Pasudotox Leading to Conformational Change and
Immunotoxin Activity Loss
X. Lu, S. Lin, N. De Mel, A. Parupudi, M. Pandey, J. Delmar, X. Wang, and J. Wang
Running Title: Deamidation impacts immunotoxin activity
1 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
ABSTRACT
Asparagine deamidation is a common posttranslational modification in which asparagine is
converted to aspartic acid or isoaspartic acid. By introducing a negative charge, deamidation
could potentially impact the binding interface and biological activities of protein therapeutics.
We identified a deamidation variant in moxetumomab pasudotox, an immunotoxin Fv fusion
protein drug derived from a 38-kilodalton (kDa) truncated Pseudomonas exotoxin A (PE38) for
the treatment of hairy-cell leukemia. Although the deamidation site, Asn-358, was outside of the
binding interface, the modification had a significant impact on the biological activity of
moxetumomab pasudotox. Surprisingly, the variant eluted earlier than its unmodified form on
anion exchange chromatography, which suggests a higher positive charge. Here we describe the
characterization of the deamidation variant with differential scanning calorimetry and hydrogen-
deuterium exchange mass spectrometry, which revealed that the Asn-358 deamidation caused the
conformational changes in the catalytic domain of the PE38 region. These results provide a
possible explanation for why the deamidation affected the biological activity of moxetumomab
pasudotox and suggest an approach that can be used for process control to ensure product quality
and process consistency.
2 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Statement of Significance
Asparagine deamidation can have a potentially significant impact on protein therapeutics.
Previous studies have revealed that deamidation at a single site significantly reduces the
biological activity of moxetumomab pasudotox, a recombinant anti-CD22 immunotoxin
developed for the treatment of B-cell malignancies. Surprisingly, despite the fact that
deamidation introduced a negative charge, the deamidated variant eluted earlier than its
unmodified form on anion exchange chromatography. In order to understand these observations,
we isolated the deamidated variant using an anion exchange column and characterized it by
differential scanning calorimetry and hydrogen-deuterium exchange mass spectrometry. The
results revealed the conformational change caused by the deamidation, which explains the
diminished biological activity of the variant and its early elution time on anion exchange
chromatography.
Keywords: Proteins Structure & Conformation; Protein Stability and Folding; Protein Post-
translational Modifications; Mass Spectrometry; Calorimetry
3 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
INTRODUCTION
Since the approval of recombinant human insulin by the U.S. Food and Drug Administration
(FDA) in 1982, the use of therapeutic proteins has grown tremendously for treatment of many
diseases and disorders, including cancer, autoimmune and inflammatory disorders, and
metabolic, cardiovascular, and infectious diseases (1, 2). Protein-based biopharmaceuticals often
have higher specificity and therefore lower potential to cause adverse effects and immune
responses than small-molecule drugs, but they also present their own challenges (2). One of the
most critical of these challenges is the chemical degradation that may occur during the
manufacturing and storage for most therapeutic proteins, which leads to one or more
posttranslational modifications (PTMs) (2, 3). Asparagine (Asn) deamidation, one of most
common nonenzymatic PTMs, has long been considered as a major degradation pathway for
many monoclonal antibodies and other therapeutic proteins (4). It is generally believed that
under neutral and basic conditions, the reaction goes through a five-membered cyclic
succinimide intermediate, which then hydrolyzes and yields both isoaspartic acid (iso-Asp) and
Aspartic acid (Asp) (5-7).
It has been reported that some PTMs could have a significant impact on the biological activity of
therapeutic proteins. For example, in a study of dornase alpha, a recombinant enzyme used for
the treatment of cystic fibrosis, deamidation of a single Asn residue resulted in loss of its
biological activity (8). Deamidation was also reported to compromise antibody colloidal stability
and to increase aggregation under low-pH conditions (9). Especially in light of the relationship
4 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
between protein structure and function, detailed structural characterization is essential to
understand the molecular basis of such reactions.
Hydrogen-deuterium exchange (HDX) mass spectrometry (MS) has emerged as a useful tool to
monitor the conformation and dynamics of therapeutic proteins (10-13). Conventional
biophysical techniques such as circular dichroism and fluorescence (14-16) provide only general
information, and it is difficult to detect subtle conformational changes or changes related only to
specific regions of proteins with these techniques. In contrast, HDX-MS can be used to detect
and localize the impact of PTMs on the structure of proteins at the peptide level, or even at the
residue-level. For example, HDX has been used to study Asp isomerization and methionine
oxidation in complementarity-determining regions (CDR) of antibodies. Although they are only
one amino acid apart in sequence, these two modifications have been found to result in opposing
conformational impacts on local and nearby regions, leading directly to different alterations on
antibody-antigen binding affinity (17). It has also been reported that CDR oxidation of
tryptophan significantly reduces antigen binding affinity, and HDX-MS data have revealed that
the modifications increased the flexibility of variable regions and disrupted local conformation
(18). While a similar experiment with Asn deamidation revealed no measurable local
conformational changes caused by the modification, this finding may have been compromised by
the low deamidation levels (~30%) in the material used in the study (19).
A few studies have explored how Asn deamidation affects the 3-dimensional structure of
proteins (20-25). Deamidation of human norovirus capsid protein has been reported to attenuate
glycan binding, and HDX-MS data have revealed localized alterations in protein flexibility,
5 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
supporting findings from nuclear magnetic resonance and X-ray crystallography studies (26). In
another HDX-MS study, deamidation was reported to slow Curli amyloid-protein aggregation
(11).
Moxetumomab pasudotox (formerly CAT-8015) is a recombinant immunotoxin developed for
the treatment of B-cell malignancies (27) that received FDA approval in September 2018 for the
treatment of adults with relapsed or refractory hairy-cell leukemia who have received at least 2
prior systemic therapies, including treatment with a purine nucleoside analog (28, 29). It is
composed of a 38-kDa truncated Pseudomonas exotoxin A (PE38) that is genetically fused to the
variable fragment (Fv) of an anti-CD22 antibody. When bound to CD22, a B-lymphocyte–
specific cell surface protein, it is rapidly internalized and processed. The Pseudomonas exotoxin
(PE) catalyzes the transfer of ADP ribose from nicotinamide adenine dinucleotide (NAD) to
elongation factor 2 (eEF2), which inhibits protein translation and leads to cell death.
Deamidation in the catalytic domain of PE38 at Asn-358 has previously demonstrated
significantly reduced its biological activity in cell-based bioassay (30), but it remains unclear
how a single residue modification could have such a marked impact on the cytotoxicity of the
molecule. Another puzzling observation is that the deamidated species eluted before the major
product peak on a positively charged anion exchange column, despite the fact that Asn
deamidation introduces an extra negative charge.
In this work, Asn-358–deamidated moxetumomab pasudotox was first isolated using ion
exchange chromatography (IEC). The structural impact of deamidation was carefully examined
by both differential scanning calorimetry (DSC) and HDX-MS, which showed alteration of the
6 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
conformation of the PE catalytic domain. In the comprehensive functional studies described here,
we gained a deep understanding of why the single deamidation diminished biological activity
and appeared as a basic variant, an early elution peak in anion exchange chromatography. After
evaluating the results from our stability study and the refolding process, a control strategy was
established and implemented in the commercial process to minimize the deamidation level and
maintain consistency.
MATERIALS AND METHODS
Moxetumomab pasudotox
Recombinant moxetumomab pasudotox used in this study was expressed in Escherichia coli
strains and was purified at AstraZeneca (Gaithersburg, MD) as previously described (31). All
other commercial chemicals were reagent grade or better. Peptide mapping and cell-based
bioassay methods were employed as previously described (30).
IEC fractionation
IEC fractionation was performed on a 1200 HPLC system equipped with a PL-SAX strong anion
exchange column (4.6 × 50 mm, 8 μm, 1000 Å; Agilent, Santa Clara, CA). Mobile phase A was
composed of 10 mM Tris and 1 mM ethylenediaminetetraacetic acid, pH 8.0, and mobile phase
B was composed of 10 mM Tris, 1 mM ethylenediaminetetraacetic acid, and 0.5 M NaCl, pH
8.0. For each experiment, 500 µL of moxetumomab pasudotox starting material at a
7 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
concentration of 1 mg/mL was loaded on the system and eluted with a linear gradient of up to
80% mobile phase B at a flow rate of 1.0 mL/min. The eluted protein was monitored by
ultraviolet absorbance at 280 nm, and 2 fractions (pre-peak and main peak) were collected with a
fraction collector. The fractions from multiple injections were pooled together and buffer
exchanged with 25 mM sodium phosphate, pH 7.4, to a final concentration of approximately 1
mg/mL. The purity of IEC fractionation was verified by injecting 25 μg of starting material and 5
μg of each fraction on the anion exchange column.
Differential scanning calorimetry
DSC experiments were performed with Microcal VP-Capillary DSC (Malvern Instruments,
Malvern, PA) at a concentration of approximately 1 mg/mL. A scan rate of 60°C per hour was
used with temperatures ranging from 20°C to 100°C. Results were analyzed with Origin 7.0
Microcal plug-in software (Malvern Instruments). Concentration-normalized data were fitted to a
non–2-state model with 2 peaks to calculate melting temperature and calorimetric enthalpy of the
individual domains after subtracting the heat capacity of the reference buffer.
HDX-MS analysis
HDX-MS analysis was performed using Waters HDX Technology (Waters, Milford, MA). Both
IEC pre-peak and main peak fractions were concentrated to 4 mg/mL prior to the HDX
experiment. Samples containing 2 µL each were diluted 10-fold with either H2O buffer (25 mM
sodium phosphate, pH 7.4) or D2O buffer (25 mM sodium phosphate, pD 7.4) at 25°C. After
8 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
incubation for 0 (undeuterated), 0.5, 1, 5, 10, 30, 60, and 120 minutes, the labeled samples were
quenched by adding 20 µL of ice-cold quench buffer (7.2 M guanidine, 500 mM Tris(2-
carboxyethyl) phosphine hydrochloride, pH 2.4) and were then further diluted with 120 µL of
ice-cold 0.1% formic acid solution. Immediately, 50 µL of diluted samples was digested by an
Enzymate Pepsin Column (5 µM, 2.1 × 30 mm; Waters) at 20°C. The peptic fragments were
collected with an Acquity BEH C18 VanGuard Pre-column (2.1 × 5 mm; Waters) and eluted into
an Acquity BEH C18 column (1.7 μm, 1.0 × 100 mm; Waters) at 0°C. Peptides separated on the
column were analyzed with a Xevo G2-XS QToF mass spectrometer (Waters). Data analysis was
performed as previously described (23), and 3-dimensional structures of PE38 were drawn based
on the structure 1IKQ from the Protein Data Bank.
CD22 binding assay
A surface plasmon resonance–based characterization assay was used to measure the kinetic
binding affinity (equilibrium dissociation constant [KD]) of moxetumomab pasudotox for CD22.
An anti-his tagged antibody was immobilized to 2 flow cells of a sensor chip, using a standard
amine coupling procedure. A reference surface was created by omitting the capture of CD22 on 1
flow cell. A dilution series of moxetumomab pasudotox was prepared and then injected in a
serial-flow fashion across the reference surface and the CD22 surface, followed by regeneration
of both flow cells. The data were reference flow cell subtracted and buffer blanked and then fit to
a 1:1 binding model, using Biacore Evaluation software (Precision Antibody, Columbia, MD).
This model yields an association rate constant (ka) and a dissociation rate constant (kd). The KD
9 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
for the interaction between moxetumomab pasudotox and CD22 was determined by the equation
KD = kd/ka.
Internalization assay
Kinetic imaging analysis was used to determine whether moxetumomab pasudotox IEC pre-peak
material was internalized into B-lymphoma cells. Moxetumomab pasudotox was conjugated with
Alexa Fluor 647 (Thermo Fisher Scientific, Waltham, MA) and was bound to Daudi cells. The
internalization of moxetumomab pasudotox was monitored with confocal fluorescence
microscopy. Quantitative assessment of kinetic images of the cell membrane and cytoplasm over
time demonstrated the internalization of moxetumomab pasudotox into the cytoplasm.
ADP ribosylation assay
An ADP ribosylation assay was used to evaluate the ADP ribosylation activity of moxetumomab
pasudotox in vitro. Moxetumomab pasudotox reference standard and test samples were serially
diluted in ultrapure nuclease-free water and incubated with an ADP ribosylation assay system
containing eukaryotic eEF2, NAD+, luciferase mRNA, complete amino acids mixture, RNase
inhibitor, and wheat germ extract in a 96-well plate for 2 hours at ambient temperature. After
incubation, an equal volume of Steady Glo Luciferase assay buffer and substrate (Promega,
Madison, WI) were added to each well and the plate was incubated at ambient temperature with
gentle shaking for 30 minutes before the luminescence was read on a plate reader. The amount of
luminescence detected in each well was directly proportional to the amount of translated
10 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
luciferase protein, which was inversely proportional to the amount of ADP ribosylation of eEF2
by moxetumomab pasudotox. The test sample was considered to be active if there was similar
inhibition of luciferase mRNA translation as in the moxetumomab pasudotox reference standard.
If there was no inhibition of luciferase mRNA translation, then the sample was considered to be
inactive.
RESULTS AND DISCUSSION
Separation and identification of Asn-358 deamidated variant
As shown in Figure 1a, IEC analysis of moxetumomab pasudotox starting material led to a 2-
peak profile, indicating the presence of 2 charge variants. The elution time of these variants (10.0
min and 12.5 min) were consistent with those in previous studies (21, 30), so these two IEC
peaks were termed “pre-peak” and “main peak.” As shown in Figure 1b and 1c, successful
enrichment of two variants was confirmed, and subsequent peptide mapping analysis revealed
that the pre-peak fraction was mainly Asn-358–deamidated moxetumomab pasudotox (97.5%),
whereas the main peak fraction contained only 9.6% deamidated Asn-358.
11 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Figure 1. IEC fractionation of moxetumomab pasudotox. (a) IEC profile of starting material; (b)
IEC profile of the pre-peak fraction; (c) IEC profile of the main peak fraction.
12 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
DSC study
As a technology that can monitor the thermal unfolding transitions of a protein, DSC provided
the first insight on the structural impact of Asn-358 deamidation. As shown in Figure 2 and
Table 1, DSC measurements of both the pre-peak and the main peak fractions showed 2 distinct
transitions, the first at ~44 ºC and the second at ~78 ºC. Native trypsin digestion of
moxetumomab pasudotox was used to assign the first peak to unfolding of the PE38 domains and
the second peak to the Fv domain (see Supporting Information), which is consistent with
previous findings (32). Comparison of the two DSC profiles revealed that both fractions shared
similar melting transition points for the Fv domain, but the PE38 domain of the pre-peak fraction
showed a decrease of 1.3°C in melting temperature and a 15-kcal loss in enthalpy compared with
the main peak fraction. These results suggest that Asn-358 deamidation had no effect on Fv
thermal stability but slightly destabilized the PE38 region, indicating the possible conformation
change in PE38.
13 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Figure 2. DSC thermograms of IEC pre-peak fraction (red solid line) and main peak fraction
(black dash line).
Table 1. DSC melting temperature and enthalpy of the pre-peak and main peak fractions
First transition Second transition
Sample Tm (ºC) Enthalpy (kcal) Tm (ºC) Enthalpy (kcal)
Pre-peak 43.5 122 77.5 186
Main peak 44.8 137 77.6 186
Tm = melting temperature.
HDX-MS analysis
To further localize the conformational change caused by Asn-358 deamidation, HDX-MS was
employed to analyze the two IEC fractions. A total of 182 peptic peptides were identified during
the HDX data processing, and the standard deviation for each exchange time point was
determined to be ±0.3 daltons (Da). Among all identified peptides, 14 were from the variable
domain of light chain (LC), resulting in 81% of the LC sequence coverage. As shown in Figure
3a, both fractions had superimposable exchange kinetic curves for the peptide LC 37-47,
indicating no conformational difference in this peptide region. Similar results were also observed
for all the 13 other LC peptides. To gain a global view of the LC, the differences in deuterium
incorporation between the 2 fractions were calculated for all the LC peptides (Figure 3b). It is
clear that all these differences were smaller than 0.3 Da and were insignificant, so it was
14 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
concluded that there was no structural difference between the two IEC fractions in the LC, which
is consistent with DSC melting temperature for the second thermal transition.
When the rest 168 peptides were identified to provide 96% sequence coverage of the VH and
PE38 domains, which are also refer as heavy chain (HC), their HDX-MS data were more
complex than those from the LC. First, consistent with the LC, all peptides located in the Fv
region (residues HC1–129) showed no difference between the 2 fractions; exchange kinetic
curves of peptides HC52–69 are shown as an example in Figure 3a. Second, most peptides in the
PE38 region (residues HC130–476) showed no or only minor differences (≤0.3 Da) in deuterium
uptake between the pre-peak and the main peak fractions, suggesting that no significant
conformational difference was caused by Asn-358 deamidation in most of the PE38 region.
Finally, obvious differences were observed for peptides associated with four regions (HC270–
280, HC301–315, HC334–362, and HC428–455). As shown in Figure 3a, the peptides in regions
HC267–275, HC301–307, HC341–361, and HC428–454 all exhibited increased deuterium
uptake in the pre-peak fraction compared with the main peak fraction, and all other peptides
related to these regions showed similar differences (Figure 3c). The differences at many
exchange time points were greater than 0.3 Da, and a few were as high as 2.5 Da. These results
indicated that some local conformational changes led to less protection against deuterium
exchange in these regions.
One limitation of the differences shown in Figure 3c is that the peptide length was not
considered, and sometimes longer peptides show more significant differences than short ones
because they are able to accumulate more backbone amide deuterium, also because back-
15 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
exchange can easily overwhelm the exchange on shorter peptides. To take the peptide length into
account, the differences in deuterium uptake by PE38 peptides between the 2 fractions were
divided by the total number of the exchangeable amide hydrogens in each peptide, depicted as
colors in the heat map (Figure 4). This heat map clearly highlights the four PE38 regions in
which the pre-peak fraction showed significantly higher deuterium uptake than the main peak
fraction. Overall, the HDX data suggest that Asn-358 deamidation had no impact on the Fv
region but did change local conformation in four regions in PE38. Considering that these regions
account for only ~24% of total PE38 sequence, these data are consistent with the 1.3°C
difference in the DSC melting temperature of the first transition.
16 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Figure 3. HDX analysis of pre-peak and main peak fractions. (a) Deuterium uptake kinetic
curves of selected peptides from the pre-peak fraction (red) and the main peak fraction (blue).
Difference charts of (b) light chain and (c) heavy chain summarize the difference in deuterium
uptake between the 2 fractions for all 7 exchange time points: 0.5 minute (orange), 1 minute
(red), 5 minutes (dark red), 10 minutes (cyan), 30 minutes (blue), 60 minutes (dark blue), and
120 minutes (black). The gray vertical bars represent the summed difference from all time points.
Figure 4. HDX heat map of PE38. Relative deuterium uptake differences between pre-peak and
main peak fractions are shown in color along the sequence for all 7 exchange time points: 0.5, 1,
5, 10, 30, 60, and 120 minutes.
In addition, careful examination of the sequence revealed that these HDX-defined regions
contained a total of 5 arginine residues. Because the HDX data suggested that these regions
gained more solvent accessibility due to the deamidation of Asn-358, the deamidated
17 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
moxetumomab pasudotox molecules might expose more positive charges that overcome 1
negative charge from Asn-358 deamidation; this would explain why the deamidated species
eluted before the unmodified ones on a positively charged anion exchange column.
Plausible mechanism for biological activity loss
As a truncated form of PE, PE38 contains part of the expression improvement domain (domain
Ib), the translocation domain (domain II), and the ADP ribosyltransferase enzyme domain
(domain III). The finding that all 4 HDX-defined regions were located in domain III suggested
that the deamidation of Asn-358 (Asn-495 in PE) led to the misfolding of the catalytic domain
and reduction of the biological activity of moxetumomab pasudotox. In a previous structural
study of the complex eEF2-PE, eEF2 mainly bound to 486-493, 576-580, and Arg-412 of PE
(33). When the HDX heat map and these eEF2 binding interfaces were demonstrated on the
crystal structure of PE in our study (Figure 5), it became obvious that the eEF2 binding interface
completely overlapped with the 4 HDX-defined regions 407–417, 438–452, 471–499, and 565–
592 in PE (see Supplementary Information for sequence alignment of PE and PE38).
Furthermore, 5 of 9 residues that constructed the NAD binding interface were also located in
these HDX-defined regions (Figure 5b) (34). All these results suggest that the conformational
changes resulting from Asn-358 deamidation disrupted binding of moxetumomab pasudotox to
eEF2 and NAD, which led to loss of PE38-mediated cell killing biological activity.
18 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Figure 5. Crystalline structure of PE (PDB 1IKQ). (a) HDX heat map of 0.5-minute exchange
time point; (b) binding interface of eEF2 (magenta) (33) and NAD (green) (34). The deamidation
site is shown in orange. All residues are numbered according to PE.
This proposed mechanism is supported by observations from further biological characterization
of the IEC pre-peak fraction as summarized in Table 2. The target binding results indicate that
19 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
the IEC pre-peak material is capable of binding to the soluble CD22 with similar affinity as main
peak, indicating that deamidation of PE38 at residue Asn-358 does not impact binding to the
CD22. Kinetic imaging analysis was used to determine whether moxetumomab pasudotox the
IEC pre-peak material is internalized into B-lymphoma cells. Quantitative assessment of kinetic
images of the cell membrane and cytoplasm over time demonstrated the internalization of
moxetumomab pasudotox into the cytoplasm. The IEC pre-peak material was internalized into
the cells with a similar kinetics as moxetumomab pasudotox starting material (Supplementary
figure S5). The ADP-ribosylation activity of moxetumomab pasudotox was assessed using an in
vitro luciferase translation assay. No inhibition of the luciferase mRNA translation was observed
for IEC pre-peak material, while the moxetumomab pasudotox starting material inhibited the
luciferase mRNA translation in a dose dependent manner as shown in Figure 6. These results
indicate that the IEC pre-peak material shows dramatically decreased activity to induce the ADP-
ribosylation of eEF2, as indicated by the changes on the catalyst domain of PE38, and therefore
caused the loss of cell-killing activity.
Table 2. Summary of IEC pre-peak biological characterization
Method Purpose Results
CD22 binding assay Measure binding affinity to CD22 kD=1.39 nM for Pre-peak
kD=1.38 nM for main peak
Internalization assay Measure cellular internalization Internalization observed
ADP-ribosylation assay Measure ADP ribosylation activity Inactive
Apoptosis bioassay Measure cell killing activity (relative Inactive
potency)
20 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Moxetumomab 4×10 6 Pre-Peak
3×10 6
U 6
L 2×10 R
1×10 6
0 0.001 0.01 0.1 1 Moxetumomab Pasudotox [g/mL]
Figure 6 Dose-responsive curves of ADP ribosylation activity of IEC Pre-Peak (triangle) and starting moxetumomab pasudotox (circle).
The correlation of IEC pre-peak levels and relative potency by apoptosis bioassay were linearly
plotted as described in a prior study (30), but the ratio of percent increase in deamidation to the
percent decrease in relative potency was not 1:1. A spiking study was therefore carried out in
which various ratios of purified IEC pre-peak to main peak were tested for relative potency. The
correlation between IEC pre-peak to potency was nonlinear (Figure 7), indicating that the IEC
pre-peak material was not only inactive but also appeared to competitively inhibit the biological
activity of moxetumomab pasudotox. The relationship between the IEC pre-peak and relative
potency can been expressed as:
y = a (1 – x)*(1 – bx),
21 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
where x is percent pre-peak, y is percent relative potency, a is the relative potency of the pure
main peak (percent pre-peak is 0), and b is the factor of competitive inhibition. In this case, b is
approximately 0.55. With an understanding of the correlation between pre-peak and relative
potency, an appropriate specification can be set for the pre-peak to ensure drug efficacy and
safety.
Figure 7. Correlation of IEC pre-peak ratio to relative potency.
Interestingly, the Asn-358 deamidation level did not change during the forced degradation study
at high pH. The crystalline structure of PE indicates that Asn-358 (Asn-495 in PE) is buried, so
Asn-358 deamidation can occur only during the refolding step, during which the molecule is
exposed to high pH and denaturing conditions. The deamidated molecule folded incorrectly.
22 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
With a clear understanding of the structural impact of the deamidation, process control can be
achieved by optimizing the refolding conditions and the purification process to consistently
produce high-quality product.
CONCLUSIONS
In this study, we report a conformational change caused by single asparagine deamidation based
on DSC and HDX MS data. It explained how Asn-358 deamidation impacted the biological
activity of moxetumomab pasudotox and why the deamidated variant eluted as an early peak in
anion exchange chromatography. Importantly, our results indicate that residue Asn-358 is
enveloped once the protein is folded and thus will not change during storage conditions. With
clear understanding of the structure-function relationship, the level of Asn-358 deamidation
variant is minimized and controlled consistently in the current process for producing clinical and
commercial supplies.
23 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
AUTHOR INFORMATION
Author Contributions
X.L. planned and organized all the experiments, conducted the HDX MS study and data
processing, and wrote most of the manuscript. S.L. contributed to interpretation of biological
data and contributed to the writing of the manuscript. N.D.M. performed the IEC fractionation
and contributed to the writing of the manuscript. A.P. performed DSC and contributed to the
writing of the manuscript. M.P. performed ADP ribosylation assay. J.D. updated the modeling
and contributed to the writing of the manuscript. X.W. contributed to the writing, reviewing, and
editing of the manuscript. J.W. performed peptide mapping analysis, summarized the data and
conclusions, and contributed to the writing of the manuscript. All authors have approved the final
version of the manuscript.
Acknowledgments
All authors are either employee of AstraZeneca or was employee of AstraZeneca when the work
was done. The authors acknowledge Timothy Pabst and Thomas Linke for providing the
Moxetumomab Pasudotox material and LeeAnn Machiesky, Erika Farmer for testing biological
activities, and Irma Vainshtein for internalization assay. We would like to thank Ben Niu and
Xiaoyu Chen for review and edit. Editorial support was provided by Deborah Shuman (all are
employee of AstraZeneca, Gaithersburg, MD, USA).
24 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
REFERENCES
1. Kinch, M. S., and R. H. Griesenauer. 2018. 2017 in review: FDA approvals of new
molecular entities. Drug Discov. Today. 23(8):1469-1473.
2. Leader, B., Q. J. Baca, and D. E. Golan. 2008. Protein therapeutics: a summary and
pharmacological classification. Nat. Rev. Drug. Discov. 7(1):21-39.
3. Jenkins, N. 2007. Modifications of therapeutic proteins: challenges and prospects.
Cytotechnology. 53(1-3):121-125.
4. Manning, M. C., D. K. Chou, B. M. Murphy, R. W. Payne, and D. S. Katayama. 2010.
Stability of protein pharmaceuticals: an update. Pharm. Res. 27(4):544-575.
5. Capasso, S., and S. Salvadori. 1999. Effect of the three-dimensional structure on the
deamidation reaction of ribonuclease A. J. Pept. Res. 54(5):377-382.
6. Capasso, S., L. Mazzarella, F. Sica, and A. Zagari. 1989. Deamidation via cyclic imide in
asparaginyl peptides. Pept. Res. 2(2):195-200.
7. Capasso, S., L. Mazzarella, and A. Zagari. 1991. Deamidation via cyclic imide of
asparaginyl peptides: dependence on salts, buffers and organic solvents. Pept. Res.
4(4):234-238.
8. Shire, S. J. 1996. Stability characterization and formulation development of recombinant
human deoxyribonuclease I [Pulmozyme, (dornase alpha)]. Pharm. Biotechnol. 9:393-
426.
9. Alam, M. E., G. V. Barnett, T. R. Slaney, C. G. Starr, T. K. Das, and P. M. Tessier. 2019.
Deamidation can compromise antibody colloidal stability and enhance aggregation in a
pH-dependent manner. Mol. Pharm. 16(5):1939-1949.
25 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
10. Berkowitz, S. A., J. R. Engen, J. R. Mazzeo, and G. B. Jones. 2012. Analytical tools for
characterizing biopharmaceuticals and the implications for biosimilars. Nat. Rev. Drug.
Discov. 11(7):527-540.
11. Wang, H., Q. Shu, C. Frieden, and M. L. Gross. 2017. Deamidation slows curli amyloid-
protein aggregation. Biochemistry. 56(23):2865-2872.
12. Oganesyan, I., C. Lento, and D. J. Wilson. 2018. Contemporary hydrogen deuterium
exchange mass spectrometry. Methods. 144:27-42.
13. Deng, B., C. Lento, and D. J. Wilson. 2016. Hydrogen deuterium exchange mass
spectrometry in biopharmaceutical discovery and development: a review. Anal. Chim.
Acta. 940:8-20.
14. Asomugha, C. O., R. Gupta, and O. P. Srivastava. 2011. Structural and functional roles of
deamidation of N146 and/or truncation of NH2- or COOH-termini in human alphaB-
crystallin. Mol. Vis. 17:2407-2420.
15. Barnett, G. V., G. Balakrishnan, N. Chennamsetty, L. Hoffman, J. Bongers, L. Tao, Y.
Huang, T. Slaney, T. K. Das, A. Leone, and S. R. Kar. 2019. Probing the tryptophan
environment in therapeutic proteins: implications for higher-order structure on tryptophan
oxidation. J. Pharm. Sci. 108(6):1944-1952.
16. Gupta, R., and O. P. Srivastava. 2004. Deamidation affects structural and functional
properties of human alphaA-crystallin and its oligomerization with alphaB-crystallin. J.
Biol. Chem. 279(43):44258-44269.
17. Yan, Y., H. Wei, Y. Fu, S. Jusuf, M. Zeng, R. Ludwig, S. R. Krystek, Jr., G. Chen, L. Tao,
and T. K. Das. 2016. Isomerization and oxidation in the complementarity-determining
regions of a monoclonal antibody: a study of the modification-structure-function
26 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
correlations by hydrogen-deuterium exchange mass spectrometry. Anal. Chem.
88(4):2041-2050.
18. Hageman, T., H. Wei, P. Kuehne, J. Fu, R. Ludwig, L. Tao, A. Leone, M. Zocher, and T.
K. Das. 2018. Impact of tryptophan oxidation in complementarity-determining regions of
two monoclonal antibodies on structure-function characterized by hydrogen-deuterium
exchange mass spectrometry and surface plasmon resonance. Pharm. Res. 36(1):24.
19. Zhang, A., P. Hu, P. MacGregor, Y. Xue, H. Fan, P. Suchecki, L. Olszewski, and A. Liu.
2014. Understanding the conformational impact of chemical modifications on
monoclonal antibodies with diverse sequence variation using hydrogen/deuterium
exchange mass spectrometry and structural modeling. Anal. Chem. 86(7):3468-3475.
20. Noguchi, S., Y. Satow, T. Uchida, C. Sasaki, and T. Matsuzaki. 1995. Crystal structure of
Ustilago sphaerogena ribonuclease U2 at 1.8 A resolution. Biochemistry. 34(47):15583-
15591.
21. Noguchi, S., K. Miyawaki, and Y. Satow. 1998. Succinimide and isoaspartate residues in
the crystal structures of hen egg-white lysozyme complexed with tri-N-acetylchitotriose.
J. Mol. Biol. 278(1):231-238.
22. Esposito, L., L. Vitagliano, F. Sica, G. Sorrentino, A. Zagari, and L. Mazzarella. 2000.
The ultra-high resolution crystal structure of ribonuclease A containing an isoaspartyl
residue: hydration and sterochemical analysis. J. Mol. Biol. 297(3):713-732.
23. Rester, U., M. Moser, R. Huber, and W. Bode. 2000. L-Isoaspartate 115 of porcine beta-
trypsin promotes crystallization of its complex with bdellastasin. Acta Crystallogr. D
Biol. Crystallogr. 56(Pt 5):581-588.
27 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
24. Eschenburg, S., and E. Schonbrunn. 2000. Comparative X-ray analysis of the un-liganded
fosfomycin-target murA. Proteins. 40(2):290-298.
25. Noguchi, S. 2010. Structural changes induced by the deamidation and isomerization of
asparagine revealed by the crystal structure of Ustilago sphaerogena ribonuclease U2B.
Biopolymers. 93(11):1003-1010.
26. Mallagaray, A., R. Creutznacher, J. Dulfer, P. H. O. Mayer, L. L. Grimm, J. M. Orduna,
E. Trabjerg, T. Stehle, K. D. Rand, B. S. Blaum, C. Uetrecht, and T. Peters. 2019. A post-
translational modification of human norovirus capsid protein attenuates glycan binding.
Nat. Commun. 10(1):1320.
27. Kreitman, R. J., and I. Pastan. 2011. Antibody fusion proteins: anti-CD22 recombinant
immunotoxin moxetumomab pasudotox. Clin. Cancer Res. 17(20):6398-6405.
28. Dhillon, S. 2018. Moxetumomab pasudotox: first global approval. Drugs. 78(16):1763-
1767.
29. Mohseni, M., H. Uludag, and J. M. Brandwein. 2018. Advances in biology of acute
lymphoblastic leukemia (ALL) and therapeutic implications. Am. J. Blood Res. 8(4):29-
56.
30. Linke, T., J. Feng, K. Yu, H. J. Kim, Z. Wei, Y. Wang, W. K. Wang, and A. K. Hunter.
2012. Process scale separation of an anti-CD22 immunotoxin charge variant. J.
Chromatogr. A. 1260:120-125.
31. Linke, T., M. T. Aspelund, C. Thompson, G. Xi, A. Fulton, M. Wendeler, T. M. Pabst, X.
Wang, W. K. Wang, K. Ram, and A. K. Hunter. 2014. Development and scale-up of a
commercial fed batch refolding process for an anti-CD22 two chain immunotoxin.
Biotechnol. Prog. 30(6):1380-1389.
28 bioRxiv preprint doi: https://doi.org/10.1101/761635; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
32. Ionescu, R. M., J. Vlasak, C. Price, and M. Kirchmeier. 2008. Contribution of variable
domains to the stability of humanized IgG1 monoclonal antibodies. J. Pharm. Sci.
97(4):1414-1426.
33. Jorgensen, R., A. R. Merrill, S. P. Yates, V. E. Marquez, A. L. Schwan, T. Boesen, and G.
R. Andersen. 2005. Exotoxin A-eEF2 complex structure indicates ADP ribosylation by
ribosome mimicry. Nature. 436(7053):979-984.
34. Li, M., F. Dyda, I. Benhar, I. Pastan, and D. R. Davies. 1996. Crystal structure of the
catalytic domain of Pseudomonas exotoxin A complexed with a nicotinamide adenine
dinucleotide analog: implications for the activation process and for ADP ribosylation.
Proc. Natl. Acad. Sci. U. S. A. 93(14):6902-6906.
29