Available via license: CC BY-ND 4.0
Content may be subject to copyright.
1
In vitro discovery and optimization of a human monoclonal antibody that
1
neutralizes neurotoxicity and lethality of cobra snake venom
2
3
Authors: Line Ledsgaard1†*, Andreas H. Laustsen1†*, Urska Pus1, Jack Wade1, Pedro Villar2,
4
Kim Boddum3, Peter Slavny2, Edward W. Masters2, Ana S. Arias4, Saioa Oscoz1, Daniel T.
5
Griffiths2, Alice M. Luther2, Majken Lindholm2, Rachael A. Leah2, Marie Sofie Møller1, Hanif
6
Ali5, John McCafferty2, Bruno Lomonte4, José M. Gutiérrez4, Aneesh Karatt-Vellatt2*
7
†These authors contributed equally to this work
8
9
Affiliations:
10
1Department of Biotechnology and Biotherapeutics, Technical University of Denmark; DK-2800
11
Kongens Lyngby, Denmark
12
2IONTAS Ltd.; Cambridgeshire CB22 3FT, United Kingdom
13
3Sophion Bioscience, DK-2750 Ballerup, Denmark
14
4Instituto Clodomiro Picado, Facultad de Microbiología, Universidad de Costa Rica; San Jos
15
11501-2060, Costa Rica
16
5Quadrucept Bio, Cambourne CB23 6DW, United Kingdom
17
*Corresponding author. Email: liljen@dtu.dk (L.L.); ahola@dtu.dk (A.H.L.);
18
akv@maxion.co.uk (A.K-V.)
19
20
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
2
One Sentence Summary: A recombinant human monoclonal immunoglobulin G antibody,
21
discovered and optimized using in vitro methods, was demonstrated to neutralize the lethal effect
22
of whole venom from the monocled cobra in mice via abrogation of α-neurotoxin-mediated
23
neurotoxicity.
24
Abstract:
25
The monocled cobra (Naja kaouthia) is one of the most feared snakes in Southeast Asia. It is a
26
highly dangerous species with a potent venom deriving its toxicity predominantly from abundant
27
long-chain α-neurotoxins. The only specific treatment for snakebite envenoming is antivenom,
28
which is based on animal-derived polyclonal antibodies. Despite the lifesaving importance of
29
these medicines over the past 120 years, and their ongoing role in combating snakebite disease,
30
major limitations in safety, supply consistency, and efficacy creates a need for a new generation
31
of improved treatments based on modern biotechnological techniques. Here, we describe the
32
initial discovery and subsequent optimization of a recombinant human monoclonal
33
immunoglobin G (IgG) antibody against α-cobratoxin using phage display technology. Affinity
34
maturation of the parental antibody by light chain-shuffling resulted in an 8-fold increase in
35
affinity, translating to a significant increase in in vitro neutralization potency and in vivo
36
efficacy. While the parental antibody prolonged survival of mice challenged with purified α-
37
cobratoxin, the optimized antibody prevented lethality when incubated with N. kaouthia whole
38
venom prior to intravenous injection. This study is the first to demonstrate neutralization of
39
whole snake venom by a single recombinant monoclonal antibody. Importantly, this suggests
40
that for venoms whose toxicity relies on a single predominant toxin group, such as that of N.
41
kaouthia, as little as one monoclonal antibody may be sufficient to prevent lethality, thus
42
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
3
providing a tantalizing prospect of bringing recombinant antivenoms based on human
43
monoclonal or oligoclonal antibodies to the clinic.
44
45
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
4
INTRODUCTION
46
Snakebite envenoming is a neglected tropical disease, which each year claims hundreds of
47
thousands of victims, who are either left permanently disfigured or meet an untimely death (1).
48
Asia is the continent where most bites and deaths occur. In Southern and Southeast Asia, the
49
monocled cobra (Naja kaouthia) is responsible for a large number of the recorded severe snakebite
50
cases (2, 3), which is exemplified by the fact that 34% of snakebite-related deaths in Bangladesh
51
from 1988 to 1989 were attributed to bites from either N. kaouthia or the closely related species,
52
N. naja (4). Life-threatening clinical manifestations of N. kaouthia envenomation include flaccid
53
paralysis due to the actions of abundant long-chain α-neurotoxins, which block neuromuscular
54
transmission by binding to the nicotinic acetylcholine receptor (nAChR) with high affinity, causing
55
a curare-mimetic effect (5, 6). These long-chain α-neurotoxins belong to the three-finger toxin
56
superfamily, which dominate the venom in terms of abundance and toxicity, as judged by their
57
high Toxicity Scores (6–8), and are thus the main toxin targets to be neutralized for successful
58
intervention in human snakebite cases.
59
Each year, the lack of access to affordable and effective treatment against snakebite
60
envenoming leaves thousands of victims in despair. Animal-derived antivenoms remain the
61
cornerstone of snakebite envenoming therapy (1) and are still produced by immunizing large
62
mammals, usually horses, with snake venom, followed by the purification of antibodies from the
63
blood plasma, resulting in polyclonal antibody preparations (9). Being heterologous products,
64
animal-derived antivenoms often lead to a range of adverse reactions whose incidence varies
65
depending on the product (10). Furthermore, it is estimated that only a fraction of the antibodies
66
in current antivenoms contribute to neutralization of relevant toxins. Large amounts of antivenom
67
are therefore required to treat a snakebite case, resulting in heterologous protein loads as high as
68
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
5
15 grams per treatment in severe envenoming cases (11, 12). Moreover, antivenoms targeting
69
elapid species in particular have a low content of therapeutic antibodies against low molecular
70
weight neurotoxins with high Toxicity Scores, as a discrepancy exists between the toxicity (high)
71
and the immunogenicity (low) of these toxins, leading to weaker immune responses in production
72
animals (13–16). Despite many advances within antibody technology and biotechnology, a need
73
remains for antivenoms with improved safety and efficacy (17, 18).
74
Recently, recombinant antivenoms based on oligoclonal mixtures of human
75
antibodies have been proposed as a cost-competitive alternative to current antivenoms (19–22).
76
Such recombinant antivenoms may offer safer and more efficacious snakebite envenoming therapy
77
due to their compatibility with the human immune system and the possibility of only including
78
efficacious antibodies, targeting medically relevant snake toxins, in the antivenom mixture (18).
79
Moreover, it has been demonstrated that such oligoclonal recombinant antivenoms, consisting of
80
carefully selected immunoglobulin Gs (IgGs), can be developed using phage display technology
81
(19). However, to date, no neutralizing human monoclonal IgG has been reported against α-
82
neurotoxins, which are among the medically most important toxins present in elapid venoms.
83
Additionally, no recombinant antibody has yet been reported to neutralize whole venom from any
84
animal following intravenous injection.
85
Here, we report the discovery and affinity maturation of a human monoclonal IgG
86
targeting α-cobratoxin, the medically most relevant toxin from N. kaouthia venom. Using chain-
87
shuffling, the affinity of the antibody was improved 8-fold, resulting in enhanced neutralization
88
both in vitro and in vivo. The increased affinity to α-cobratoxin improved the IgG from one that
89
was able to prolong survival of mice challenged with purified α-cobratoxin to one that could
90
neutralize the lethal effect of whole N. kaouthia venom in mice.
91
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
6
RESULTS
92
A phage display derived fully human antibody that prolongs survival in vivo
93
A naïve scFv phage display library (23) was used to carry out three rounds of panning against
94
biotinylated α-cobratoxin from N. kaouthia immobilized on a streptavidin coated surface.
95
Antibody-encoding genes (scFv format) were isolated from both the second and third panning
96
rounds and subcloned into a bacterial expression plasmid (24). In total, 282 clones harboring this
97
expression plasmid were picked for antibody expression and subjected to binding analysis by
98
DELFIA as previously described (19). Of these, 36 clones displaying specific binding signal
99
against α-cobratoxin (25 times above the background binding signal) were picked for DNA
100
sequencing and further characterization (Fig. 1A).
101
A total of 28 scFvs with unique VH and VL CDR3 regions were evaluated in an expression-
102
normalized capture (ENC) assay, which eliminates the expression variation between the clones
103
and allow ranking based on affinity. The six α-cobratoxin-binding scFvs that yielded the highest
104
binding signals (Fig. 1B) were selected for reformatting into the IgG format and transiently
105
expressed using Expi293F™ cells. All six α-cobratoxin-targeting antibodies retained binding to α-
106
cobratoxin upon conversion. However, when the antibodies and α-cobratoxin were preincubated
107
and administered i.v. to mice, the antibodies failed to prevent lethality, although they did succeed
108
in prolonging survival significantly (Fig. 1C). One plausible explanation for the limited efficacy
109
of these antibodies could be due to their sub-optimal binding affinity to α-cobratoxin.
110
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
7
111
Fig. 1. Affinity ranking of scFvs and Kaplan-Meier survival curves for mice co-
112
administered with IgG antibodies and α-cobratoxin. (A) Direct and ENC DELFIA of 36
113
monoclonal scFv-containing supernatants. (B) Direct and ENC DELFIA of the top six
114
monoclonal scFv-containing supernatants. (C) Kaplan-Meier survival curves for mice co-
115
administered with IgG antibodies and α-cobratoxin. α-cobratoxin refers to mice injected with α-
116
cobratoxin alone.
117
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
8
Affinity maturation using chain-shuffling
118
In vitro affinity maturation strategies involve two key steps, diversification of the primary
119
antibody sequence and enrichment of affinity-improved antibody variants using a selection
120
platform such as phage display technology. Diversification of primary antibody sequence can be
121
achieved by introducing mutations to the variable regions using random or targeted mutagenesis.
122
Alternatively, new combinations of heavy and light variable regions can be made by
123
recombining selected heavy or light chains with a repertoire of partner chains by a process
124
known as chain shuffling (25). Here, the 368_01_C01 antibody was chosen for further study due
125
to its combined performance in the ENC DEFLIA and in vivo experiments. This antibody was
126
therefore subjected to light chain shuffling to diversify sequence by pairing the VH chain with a
127
library of naïve kappa and lambda light chains. Following library generation, three rounds of
128
stringent panning against α-cobratoxin were completed. For precise control of antigen
129
concentration, phage display selections were carried out in solution phase. The phage antibodies
130
were allowed to bind to biotinylated α-cobratoxin in solution, and the bound phage was
131
subsequently captured using streptavidin-coated beads for washing and elution. The antigen
132
concentration was lowered 10-50-fold in each round to selectively enrich antibodies with high
133
affinity. The polyclonal phage outputs for the selections were tested for binding to α-cobratoxin
134
through polyclonal DELFIA, revealing that α-cobratoxin-binding scFvs were present in all three
135
rounds, while negligible binding to streptavidin was detected, see Fig. 2A. Then, scFv genes
136
from the third panning rounds were subcloned into the bacterial expression vector. A total of 184
137
monoclonal colonies from each of the two third-round selections were picked, and the soluble
138
scFvs were assessed for binding to α-cobratoxin, revealing 290 out of the 368 (79%) displaying a
139
binding signal against the antigen, see Fig. 2B. Among the 290 hits, 60 clones were randomly
140
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
9
picked for further characterization, including an ENC DELFIA and DNA sequencing, see Fig.
141
2C for ENC data. DNA sequencing revealed that 14 of the 60 scFvs had unique CDRL3
142
sequences. The 13 most promising of these scFvs, based on their ranking in the ENC assay, were
143
converted to the IgG1 format and expressed in Expi293F™ cells. A direct DELFIA assay was
144
used to confirm that all 13 antibodies retained binding in the IgG1 format. An ENC DELFIA was
145
employed to rank the binding of the IgGs, and along with their expression yields, 2552_02_B02
146
was picked as the top candidate, see Fig. 2D.
147
To assess the improvement in affinity derived from the exchange of the parental antibody
148
light chain, Surface Plasmon Resonance (SPR) was employed. Here, the parent (368_01_C01)
149
and 2552_02_B02 were reformatted as monovalent Fabs and used for the affinity measurements
150
to avoid avidity effects. For the parent clone, the affinity (equilibrium dissociation constant, KD)
151
was determined to 3.85 nM, whereas the matured clone displayed an affinity of 490 pM,
152
representing an approximate 8-fold increase in affinity. This increase in affinity resulted from the
153
improvement of both the on and off-rate of the original antibody, 2.4-fold and 3.3-fold,
154
respectively (Fig. 2 E and F).
155
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
10
156
Fig. 2. Binding, expression, and binding kinetics characterization of discovered antibodies.
157
(A) Polyclonal phage DELFIA of selections outputs from three selection rounds (where antigen
158
concentration was lowered 10-50-fold between each round) showing binding to α-cobratoxin and
159
no binding to streptavidin. (B) Direct monoclonal DELFIA of 384 monoclonal scFv-containing
160
supernatants. (C) Direct monoclonal DELFIA signals (in black on the left Y-axis) and ENC
161
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
11
DELFIA signals (in blue on the right Y-axis) of the 60 monoclonal scFvs selected for
162
sequencing. (D) ENC DELFIA signals of IgG-containing supernatants of the 13 clones that were
163
converted to the IgG format. 2552_02_B02 was selected for further characterization. For (B),
164
(C), and (D), data from the parent scFv (368_01_C01) is shown furthest to the right. (E) 1:1
165
binding model (black lines) fitted to the SPR data kinetics of parental antibody (368_01_C01) in
166
the Fab format. (F) 1:1 binding model (black lines) fitted to the SPR data of the affinity matured
167
clone (2552_02_B02) in the Fab format.
168
Affinity matured IgGs show potent blocking of toxin:receptor binding interaction
169
To assess if the increase in affinity achieved from affinity maturation translated to improved
170
blocking of the binding interaction between the AChR and α-cobratoxin, a receptor blocking
171
assay was performed. A schematic representation of the assay can be seen in Fig. 3 A.
172
Results revealed that both IgGs were able to fully abrogate the binding between the
173
receptor subunit and α-cobratoxin, though, the affinity-matured clone was able to prevent the
174
binding at lower concentrations than the parent antibody. The IC50 values were 0.32 nM (CI95
175
0.28-0.36 nM) and 1.38 nM (CI95 1.02-1.86 nM) for 2552_02_B02 and 368_01_C01,
176
respectively (Fig. 3 B). The 8.1-fold increase in affinity to α-cobratoxin therefore resulted in a
177
4.3-fold improvement of IC50 in this blocking assay.
178
179
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
12
180
Fig. 3. Inhibition of the binding interaction between the α7-AChR and α-cobratoxin using
181
IgGs. (A) Schematic representation of a receptor blocking assay, wherein the ability of IgG
182
antibodies to inhibit the interaction between the α7 subunit of AChR (α7-AChR) and α-
183
cobratoxin can be quantified. (B) Antibodies block α-cobratoxin binding to its receptor (α7-
184
AChR) in a concentration-dependent manner. As a negative control an IgG specific to
185
dendrotoxins was used and showed no blocking.
186
Increased neutralization potency in in vitro neutralization assay
187
To assess if the increase in affinity and ability to block the α7-AChR:α-cobratoxin interaction for
188
the affinity matured antibody also resulted in an improved ability of the antibody to protect
189
nAChR function, functional neutralization assays were conducted using automated patch-clamp
190
electrophysiology. First, the EC80 value for ACh was established (70 µM, Fig. 4 A and B), and
191
the IC80 for α-cobratoxin was determined (4 nM, Fig. 4 C and D). Then, titrated 368_01_C01 and
192
2552_02_B02 were preincubated with α-cobratoxin and tested to examine the ability of the
193
antibodies to neutralize the current-inhibiting activity of α-cobratoxin. As a negative control, a
194
dendrotoxin-binding IgG was included. This irrelevant IgG showed no effects, while the α-
195
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
13
cobratoxin-recognizing antibodies were able to fully abrogate α-cobratoxin activity.
196
Furthermore, the affinity-matured clone, 2552_02_B02, was a more potent neutralizer with an
197
EC50 of 2.6 nM (CI95 2.3-2.9 nM), while the parental clone (368_01_C01) exhibited an EC50 of
198
8.1 nM (CI95 6.6-10.0 nM). Relative to the concentration of α-cobratoxin used, these data
199
indicate that 0.65 IgG molecules were needed per toxin molecule for 50% neutralization for
200
2552_02_B02, whereas 2.03 IgG molecules were needed per toxin to achieve the same effect
201
with 368_01_C01. Hence, the increase in affinity between the 2552_02_B02 clone and α-
202
cobratoxin resulted in increased functional neutralization in vitro.
203
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
14
204
Fig. 4. In vitro neutralization of inhibition of nAChR by α-cobratoxin. Automated patch-
205
clamp experiments conducted using a QPatch (Sophion Bioscience). (A) Sweep plot and (B)
206
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
15
concentration-response curve showing the relationship between increased ACh concentration and
207
the measured current running across the cell membrane. 70 µM ACh was used throughout the
208
rest of the experiments. (C) Sweep plot and (D) concentration-response curve showing how
209
increasing concentrations of α-cobratoxin result in a decrease in the current measured. 4 nM α-
210
cobratoxin was used, resulting in approximately 80% inhibition of the current. (E) Sweep plot
211
(368_01_C01) and (F) concentration-response curves showing how increasing concentrations of
212
the two IgGs preincubated with α-cobratoxin result in better protection of the nAChR, as the loss
213
of current mediated by α-cobratoxin is prevented in a dose-dependent manner. An irrelevant IgG
214
was used as a control, which did not prevent the inhibitory effects of α-cobratoxin.
215
Affinity matured antibody neutralizes Naja kaouthia whole venom in vivo
216
To determine the ability of 2552_02_B02 to neutralize the effects of N. kaouthia whole venom, a
217
mouse lethality assay was set up, as this is the gold standard of the WHO for the assessment of
218
the preclinical efficacy of antivenoms. 2 LD50s of N. kaouthia whole venom were preincubated
219
with the IgG in different molar ratios for 30 minutes before injecting the mixture i.v. in mice. As
220
controls, venom was injected alone, in combination with a negative (irrelevant) control IgG
221
(Nivolumab), and in combination with a commercial polyclonal antivenom as a positive control.
222
Survival was recorded and is illustrated in Fig. 5. Mice injected with either venom alone or in
223
combination with Nivolumab all died within 30 minutes after injection with signs of
224
neuromuscular paralysis (Fig. 5 A). At 48 hours post injection, mice receiving venom and
225
polyclonal antivenom were alive and showed no signs of neurotoxicity, at which point the study
226
on this group was concluded. For the groups of 1:1 and 1:2 toxin:IgG dosages, the surviving
227
mice at 24 hours were showing clear signs of toxicity, and it was decided to extend the
228
observation period until 48 hours owing to the novel nature of these antibodies and the need to
229
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
16
know whether the neutralization observed in the first hours could be reverted. At the 48-hour
230
time point, all mice in the 1:1 ratio were dead, and only one mouse in the 1:2 toxin:IgG dosage
231
group remained alive. Therefore, a new experiment increasing the dosage to 1:4 toxin:IgG was
232
set up. For this group, all mice survived the 24-hour period, one mouse was dead at 48 hours, and
233
by 72 hours post injection no further deaths had occurred (Fig 5 B). These results clearly
234
demonstrate that the therapeutic effect of 2552_02_B02 on mice injected with a lethal amount of
235
N. kaouthia whole venom were dose-dependent in vivo, as expected for specific antibody
236
therapeutics. Even further, at an α-neurotoxin to IgG molar ratio of 1:4, all mice survived the
237
observation period recommended by the WHO for this type of study (24 hours), and even at the
238
72-hour mark, three out of four mice had survived.
239
240
Fig. 5. Kaplan-Meier curves showing survival of mice co-administered with N. kaouthia
241
whole venom and IgG 2552_02_B02. (A) Illustrates that the death of the venom only and
242
Nivolumab control groups occurred within 30 minutes after injection, whereas those receiving
243
venom and antivenom survived the 48 hour observation period. (B) Illustrates the survival of the
244
groups of mice receiving IgG and venom. 2552_02_B02 in 1:1 and 1:2 toxin:IgG molar ratios
245
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
17
were observed for 48 hours, whereas mice receiving the 1:4 molar ratio were observed for 72
246
hours.
247
DISCUSSION
248
Previously, we described the discovery of an oligoclonal mixture of human antibodies capable of
249
neutralizing dendrotoxin-mediated neurotoxicity of black mamba venom in a rodent model (19).
250
Although the cocktail of antibodies tested in that study did neutralize whole venom, the model,
251
using intracerebroventricular injection (i.c.v.), did not account for the effects elicited by α-
252
neurotoxins, since their main target is the nAChR in the neuromuscular junctions. Thus the i.c.v.
253
model is not nearly as clinically relevant as i.v. injection, which is recommended by WHO as the
254
standard for assessing antivenoms. Another study has reported in vivo neutralization of α-
255
cobratoxin-induced lethality by a VHH and a VHH2-Fc following intraperitoneal injection in
256
mice (26). However, to date no study has successfully demonstrated the neutralization of
257
lethality caused by a whole venom (or a purified toxin) preincubated with a recombinant
258
monoclonal IgG antibody following i.v. injection.
259
In this study, we demonstrate that a recombinant human monoclonal IgG antibody, discovered
260
and optimized entirely in vitro by phage display technology, was able to neutralize lethality in
261
mice challenged i.v. with whole venom from N. kaouthia. This clearly showcases the utility of in
262
vitro selection methods for the discovery of efficacious antivenom antibodies against animal
263
toxins with reduced immunogenicity, which may be challenging for traditional in vivo based
264
discovery approaches (27). Moreover, this study has also elucidated the mechanism of action of
265
the neutralizing antibody using receptor blocking and automated patch clamp electrophysiology
266
assays, which revealed that the antibody could abrogate neurotoxicity by preventing the
267
medically most important toxin, α-cobratoxin, from interacting with the nAChR. This first report
268
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
18
of a recombinant monoclonal antibody neutralizing whole venom from a snake following i.v.
269
injection thus presents an integrated approach for future discovery and evaluation of recombinant
270
antibodies against toxins from snake and other animal venoms. In this relation, it is, however,
271
important to note that the neutralization of this particular venom by a single IgG cannot be
272
extrapolated to all other snake venoms due to their complex composition of different medically
273
relevant toxin families that each may require one or more antibodies for neutralization (28). In
274
many cases, co-administration with other antibodies or small molecule inhibitors, such as
275
varespladib, batimastat, or marimastat (29, 30), might be necessary to achieve full protection
276
(18). Furthermore, the biophysical and pharmacokinetic properties of the IgG reported in this
277
study have not been investigated and therefore these properties remain unknown. Antibody
278
pharmacokinetics play a significant role in drug efficacy (31), while antibody biophysics can
279
have major influence on antibody manufacturability and stability (32–34). Additional
280
investigation of these properties is therefore warranted prior to further preclinical and clinical
281
assessments.
282
Finally, a regulatory uncertainty currently exists regarding whether recombinant antivenoms
283
based on monoclonal or oligoclonal antibodies will be regulated as blood products, similar to
284
existing plasma-derived antivenoms, or whether recombinant antivenoms will be viewed as
285
biotherapeutic products to be regulated as biopharmaceuticals by relevant authorities.
286
Establishment of a regulatory framework for recombinant antivenom products is thus a necessity
287
for bringing such new snakebite envenoming therapies swiftly to the clinic.
288
Nonetheless, the advances in the discovery, optimization, and assessment of monoclonal
289
antibodies against snake toxins described in this study represent an important technical milestone
290
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
19
towards the application of in vitro developed recombinant antivenoms as a therapeutic
291
intervention in snakebite envenoming in the future.
292
MATERIALS AND METHODS
293
Study design
294
The original objective of the study was to discover human monoclonal IgGs that could neutralize
295
α-cobratoxin in vivo. When proven unsuccessful, the aim of the study became to test if affinity
296
maturation using chain-shuffling of an α-cobratoxin-specific IgG antibody would result in
297
improved in vitro and in vivo neutralization and to test if administration of only α-cobratoxin
298
specific IgGs would enable full neutralization of N. kaouthia whole venom in vivo. The initial
299
discovery and subsequent affinity maturation of α-cobratoxin specific antibodies was carried out
300
in vitro. The following assessment of binding was conducted on the antibodies in different
301
formats (scFv, Fab, and IgG) using several different assays (direct DELFIA, ENC DELFIA, and
302
SPR). The following assessment of in vitro neutralization was conducted using two different
303
methods (automated patch-clamp and receptor blocking) at different research institutions, both
304
reaching similar conclusions that the IgGs were neutralizing. For the in vivo neutralization of
305
lethality assays, mice were randomized into treatment groups of 3-4 mice per group (group size
306
determined based on previous studies) with predefined endpoint at 24 hours; in some cases,
307
owing to the novel nature of the antibodies, the endpoint of these in vivo assays was extended.
308
All assays were conducted with 2-5 repeats (except for animal experiments) in at least technical
309
triplicates with negative controls included in all experiments and positive controls included when
310
possible.
311
Toxin preparation
312
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
20
α-cobratoxin was obtained in lyophilized form from Latoxan SAS, France. The toxin was
313
reconstituted in phosphate buffered saline (PBS) and biotinylated using a 1:1 (toxin:biotinylation
314
reagent) molar ratio as previously described (19). Following biotinylation, Amicon® Ultra-4
315
Centrifugal Filter Units with a 3 kDa membrane were used for purification of the biotinylated
316
toxin. Purification was performed at 8 oC and consisted of three washes of 4 mL PBS. The
317
protein concentration was measured by the absorbance at 280 nm using a Nanodrop and adjusted
318
using the extinction coefficient. The degree of biotinylation was analyzed using MALDI-TOF in
319
a Proteomics Analyzer 4800 Plus mass spectrometer (Applied Biosystems).
320
Initial discovery and assessment of parent clone
321
The initial discovery of the parent antibody clone 368_01_C01 was performed by panning the
322
IONTAS phage display library (diversity of 4 × 1010 human scFv clones) against biotinylated α-
323
cobratoxin captured by streptavidin in a Maxisorp vial, followed by subcloning and expression of
324
scFv genes in BL21(DE3) E. coli and DELFIA-based screening of the scFv-containing
325
supernatants as previously described (19). Thirty-eight binding clones were cherry-picked and
326
sequenced (Eurofin Genomics sequencing service) using the S10b primer
327
(GGCTTTGTTAGCAGCCGGATCTCA). The binding strengths of the clones were ranked using
328
an expression-normalized capture (ENC) assay, and the top six scFvs displaying the highest
329
binding signals were reformatted into the IgG format and expressed in Expi293™ cells (Thermo
330
Fisher) and subsequently purified using an Äkta Pure system (GE Healthcare) as previously
331
described (19). The functionality of purified IgGs was confirmed using a DELFIA-based binding
332
assay (19).
333
Library generation using chain-shuffling
334
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
21
The VH region of antibody 368_01_C01 was PCR amplified from pINT3 plasmid DNA using
335
pINT3 Nco FWD (TCTCTCCACAGGCGCCATGG) and IgG1 CH1 Xho Rev
336
(CCCTTGGTGGAGGCACTCGAG) primers using Platinum™ SuperFi II Green PCR Master
337
Mix (Invitrogen, 12369010). The PCR product was cloned into pIONTAS1 (23) vector harboring
338
the naïve VL lambda and kappa chain libraries using NcoI and XhoI restriction endonucleases.
339
Ligation reactions were carried out for 16 hours at 16 oC and contained 160 ng of insert and 400
340
ng of vector DNA in a total volume of 40 µL. Ligations were purified using the MinElute PCR
341
Purification Kit (Qiagen, 28004), and eluted in 10 µL nuclease free water. The purified ligation
342
product was transformed into 200 µL of electrocompetent TG1 cells (Lucigen, 60000-PQ763-F)
343
followed by addition of 6 mL of recovery medium (Lucigen, F98226-1) and incubation at 37 oC
344
for one hour at 280 rpm rotation. Cells were plated on 2xTY agar plates supplemented with 100
345
µg/mL ampicillin and 2% glucose. Dilutions of the transformations were also plated to determine
346
library size, which was 1.01 × 108 for the lambda library and 1.67 × 108 for the kappa library with
347
more than 96% of the transformants being positive for insertion of heavy chain insert, as
348
determined by colony PCR.
349
Library rescue and solution-based phage display selection
350
Rescue of phages from the chain-shuffled libraries and the three rounds of selections were
351
performed as described elsewhere (23), except that the phages were not concentrated using PEG
352
precipitation, but phage-containing supernatants were used directly for selections. Deselection of
353
streptavidin-specific phages was performed before each round of selection using 80 µL of
354
streptavidin-coated Dynabeads (Invitrogen, M-280). Additionally, the selections were conducted
355
using biotinylated α-cobratoxin that was captured using 80 µL of streptavidin-coated Dynabeads
356
(Invitrogen, M-280). The concentration of α-cobratoxin was decreased through the three rounds
357
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
22
of selections starting at 10 nM in the first round and ending at 20 pM in the third round. The
358
kappa and lambda libraries were mixed before the first round of selections.
359
Subcloning, primary screening, and sequencing of scFvs
360
Subcloning of the α-cobratoxin binding selection output into pSANG10-3F and primary
361
screening of candidates was performed as described elsewhere (19). In brief, scFv genes from the
362
selection outputs were subcloned from the pIONTAS1 phagemid vector to the pSANG10-3F
363
expression vector using NcoI and NotI restriction endonuclease sites and transformed into E. coli
364
strain BL21 (DE3) (New England Biolabs). From each of the two subcloned selection outputs,
365
184 colonies were picked and expressed in 96 well plates. The scFvs were assessed for their
366
binding to biotinylated α-cobratoxin (5 µg/mL) indirectly immobilized on black Maxisorp plates
367
(Nunc) with streptavidin (10 µg/mL) using a DELFIA-based assay. In total, 60 clones binding to
368
α-cobratoxin were cherry-picked and sequenced (Eurofin Genomics sequencing service) using
369
S10b primer (GGCTTTGTTAGCAGCCGGATCTCA). The antibody framework and CDR
370
regions were annotated, and light chain CDR3 regions were used to identify 14 unique clones.
371
IgG expression and purification
372
VH and VL genes of 13 unique α-cobratoxin-binding scFvs were converted to the IgG1 format as
373
previously described (19). The binding of the IgGs was confirmed and ranked using and
374
expression-normalized capture (ENC) assay. Briefly, black Maxisorp plates (Nunc) were coated
375
overnight with an anti-human IgG (Jackson ImmunoResearch, 109-005-098). Plates were
376
washed thrice with PBS and blocked with PBS supplemented with 3% milk protein. Plates were
377
washed thrice with PBS and 0.25x unpurified IgG-containing culture supernatant in PBS
378
supplemented with 3% milk protein was added before incubating for one hour at room
379
temperature. Plates were washed thrice with PBS-T and thrice with PBS before adding either 1
380
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
23
nM or 100 pM biotinylated α-cobratoxin in PBS supplemented with 3% milk protein to each
381
well. After one hour of incubation, the plates were washed thrice with PBS-T and thrice with
382
PBS. Then, 1 µg/mL of Europium-labeled Streptavidin (Perkin Elmer, 1244- 360) in DELFIA
383
Assay Buffer (Perkin Elmer, 4002-0010) was added. Following 30 minutes of incubation, plates
384
were washed thrice with PBS-T and thrice with PBS, and DELFIA Enhancement Solution
385
(Perkin Elmer, 4001-0010) was added for detection of binding. Based on these results, the two
386
top clone, 2552_02_B02, was expressed and purified as described previously (19).
387
Fab expression and purification
388
VH and VL genes of 13 unique α-cobratoxin-binding scFvs were converted to the Fab format as
389
performed for IgGs as described previously (19), except the Fab-vector, pINT12, was used
390
instead of the pINT3 IgG1 vector.
391
Surface plasmon resonance
392
The binding affinity of the discovered antibodies for α-cobratoxin was determined using Surface
393
Plasmon Resonance (SPR) (BIAcore T100, GE Healthcare). All measurements were performed
394
at 25 oC using 10 mM HEPES, 150 mM NaCl, and 3 mM EDTA at pH 7.4 as running buffer.
395
Immobilisation of α-cobratoxin on CM5 sensor chips (Cytiva, BR100530) was performed by
396
amine coupling using 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC)/ N-
397
hydroxysuccinimide (NHS) surface activation followed by injection of 5 µg/mL of α-cobratoxin
398
in 10 mM NaOAc pH 4 to btain a final immobilization level of 23 response units (RU). The
399
sensor chip was inactivated using ethanolamine. The Fabs in concentrations ranging from 81 nM
400
to 390 pM were injected at 40 µL/minute for 120 seconds and dissociation was recorded for 450
401
seconds. Following dissociation, the sensor was regenerated using two injections (15-20
402
seconds) of 20 mM NaOH. Measurements were conducted using 5-7 analyte concentrations for
403
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
24
each antibody. The blank subtracted data was analyzed using the BIAcore T100 Evaluation
404
Software employing a 1:1 Langmuir binding model.
405
Receptor blocking DELFIA
406
The receptor blocking assay was adapted from Ratanabanangkoon et al. (35). Black Maxisorp
407
plates (Nunc) were coated overnight with 100 µL of 5 µg/mL human α7-acetycholine receptor
408
chimera (adapted from (36)) in PBS. Plates were washed thrice with PBS and blocked with 1%
409
BSA in PBS. 4 nM of biotinylated α-cobratoxin with various concentrations of 368_01_C01,
410
2552_02_B02, or a negative control IgG specific to dendrotoxins in 0.1% BSA was prepared and
411
preincubated for 30 minutes at room temperature. Plates were washed thrice, and 100 µL of the
412
preincubated toxin and antibody mixture was added to the blocked wells. Following incubation
413
for one hour, the plates were washed thrice with PBS-T (PBS, 0.1% Tween-20) and thrice with
414
PBS, and 100 µL of 1 µg/mL of Europium-labeled Streptavidin (Perkin Elmer, 1244-360) in
415
0.1% BSA was added. Following 30 minutes of incubation, plates were washed thrice with PBS-
416
T and thrice with PBS and 100 µL of DELFIA Enhancement Solution (Perkin Elmer, 4001-
417
0010) was added to each well. Signals were measured using a VICTOR Nivo Multimode
418
Microplate Reader using excitation at 320 nm and emission at 615 nm. Each antibody
419
concentration was tested in quadruplicate.
420
Electrophysiology
421
Planar whole-cell patch-clamp experiments were carried out on a QPatch II automated
422
electrophysiology platform (Sophion Bioscience), where 48-channel patch chips with 10 parallel
423
patch holes per channel (patch hole diameter ∼1 μm, resistance 2.00 ± 0.02 MΩ) were used.
424
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
25
The cell line used was a human-derived Rhabdomyosarcoma RD cell line (CCL-136, from
425
ATCC), endogenously expressing the muscle-type nicotinic acetylcholine receptors (nAChR),
426
composed of the α1, β1, δ, γ, and ε subunits. The cells were cultured according to the
427
manufacturer’s guideline, and on the day of the experiment, enzymatically detached from the
428
culture flask and brought into suspension.
429
For patching, the extracellular solution contained: 145 mM NaCl, 10 mM HEPES, 4 mM KCl, 1
430
mM MgCl2, 2 mM CaCl2, and 10 mM glucose, pH adjusted to 7.4 and osmolality adjusted to 296
431
mOsm. The intracellular solution contained: 140 mM CsF, 10 mM HEPES, 10 mM NaCl, 10
432
mM EGTA, pH adjusted to 7.3, and osmolality adjusted to 290 mOsm.
433
In the experiments, an nAChR-mediated current was elicited by 70 µM acetylcholine (ACh,
434
Sigma-Aldrich), approximately the EC80 value, and after compound wash-out, 2 U
435
acetylcholinesterase (Sigma-Aldrich) was added to ensure complete ACh removal. The ACh
436
response was allowed to stabilize over three ACh additions, before the fourth addition was used
437
to evaluate the effect of α-cobratoxin (4 nM α-cobratoxin, reducing the ACh response by 80%),
438
preincubated with varying concentrations of IgGs. α-cobratoxin and IgGs were preincubated at
439
room temperature for at least 30 minutes before application, and the patched cells were
440
preincubated with α-cobratoxin and IgG for 5 minutes prior to the fourth ACh addition. As a
441
negative control, an IgG specific to dendrotoxins was included.
442
The inhibitory effect of α-cobratoxin was normalized to the full ACh response (fourth response
443
normalized to third response), plotted in a non-cumulative concentration-response plot, and a
444
Hill fit was used to obtain EC50 values for each IgG. The data analysis was performed in Sophion
445
Analyzer (Sophion Bioscience) and GraphPad Prism (GraphPad Software).
446
Animals
447
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
26
In vivo assays were conducted in CD-1 mice of both sexes of 18-20 g body weight, supplied by
448
Instituto Clodomiro Picado, following protocols approved by the Institutional Committee for the
449
Use and Care of Animals (CICUA), University of Costa Rica (approval number CICUA 82-08).
450
Mice were housed in cages in groups of 3-4 and were provided food and water ad libitum.
451
In vivo preincubation experiments
452
The neutralization activity of the naïve IgGs against α-cobratoxin was tested by i.v. injection in
453
groups of three mice. 4 µg of α-cobratoxin (corresponding to 2 LD50s) and 150 µg of
454
corresponding IgG (α-cobratoxin:IgG = 1:2 molar ratio) were dissolved in PBS, preincubated (30
455
min at 37 °C), and injected in the caudal vein, using an injection volume of 100 µL. Control mice
456
were injected with either anti-lysozyme IgG and α-cobratoxin or α-cobratoxin alone. Deaths
457
were recorded and Kaplan-Meier curves were used to represent mouse survival along time.
458
For the affinity matured clone, similar in vivo experiments were conducted, except the
459
IgG was preincubated with 9.12 µg of N. kaouthia whole venom (corresponding to 2 LD50s) at a
460
1:1, 1:2, and 1:4 α-neurotoxin:IgG molar ratio. For calculating molar ratios, it was estimated that
461
55% of N. kaouthia venom consist of α-neurotoxins, based on a toxicovenomic study of the
462
venom (7). All injections were performed as described above on groups of four mice. Control
463
mice were injected with either Nivolumab (irrelevant IgG control) incubated with N. kaouthia
464
venom or N. kaouthia venom alone. As a positive control for N. kaouthia venom neutralization,
465
Snake Venom Antiserum from VINS Bioproducts Limited (Batch number: 01AS13100) was
466
used. According to the manufacturer, the potency of this antivenom against the venom of Naja
467
naja is 0.6 mg venom neutralized per mL antivenom. Since no information is provided on the
468
neutralization of N. kaouthia venom, we used a ratio of 0.2 mg venom per mL antivenom to
469
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
27
ensure neutralization. Survival was monitored for 24-72 hours, and results are presented in
470
Kaplan-Meier curves.
471
472
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
28
References
473
1. J. M. Gutiérrez, J. J. Calvete, A. G. Habib, R. A. Harrison, D. J. Williams, D. A. Warrell,
474
Snakebite envenoming, Nat. Rev. Dis. Primer 3, 1–21 (2017).
475
2. C. Viravan, U. Veeravat, M. J. Warrell, R. D. G. Theakston, D. A. Warrell, ELISA
476
Confirmation of Acute and Past Envenoming by the Monocellate Thai Cobra (Naja kaouthia),
477
Am. J. Trop. Med. Hyg. 35, 173–181 (1986).
478
3. J. Meier, J. White, Eds., Handbook of: Clinical Toxicology of Animal Venoms and Poisons
479
(CRC Press, Boca Raton, 1995).
480
4. R. O. for S.-E. Asia, W. H. Organization, Guidelines for the management of snakebites (WHO
481
Regional Office for South-East Asia, 2016; https://apps.who.int/iris/handle/10665/249547).
482
5. Handbook of clinical toxicology of animal venoms and poisons: Meier, J. and White, J. (eds),
483
752 pp. Boca Raton, FL: CRC Press (1995), Toxicon 35, 617 (1997).
484
6. M. Alkondon, E. X. Albuquerque, alpha-Cobratoxin blocks the nicotinic acetylcholine
485
receptor in rat hippocampal neurons, Eur. J. Pharmacol. 191, 505–506 (1990).
486
7. A. H. Laustsen, J. M. Gutiérrez, B. Lohse, A. R. Rasmussen, J. Fernández, C. Milbo, B.
487
Lomonte, Snake venomics of monocled cobra (Naja kaouthia) and investigation of human IgG
488
response against venom toxins, Toxicon Off. J. Int. Soc. Toxinology 99, 23–35 (2015).
489
8. A. H. Laustsen, B. Lohse, B. Lomonte, M. Engmark, J. M. Gutiérrez, Selecting key toxins for
490
focused development of elapid snake antivenoms and inhibitors guided by a Toxicity Score,
491
Toxicon Off. J. Int. Soc. Toxinology 104, 43–45 (2015).
492
9. G. León, M. Vargas, Á. Segura, M. Herrera, M. Villalta, A. Sánchez, G. Solano, A. Gómez,
493
M. Sánchez, R. Estrada, J. M. Gutiérrez, Current technology for the industrial manufacture of
494
snake antivenoms, Toxicon Off. J. Int. Soc. Toxinology 151, 63–73 (2018).
495
10. G. León, M. Herrera, Á. Segura, M. Villalta, M. Vargas, J. M. Gutiérrez, Pathogenic
496
mechanisms underlying adverse reactions induced by intravenous administration of snake
497
antivenoms, Toxicon Off. J. Int. Soc. Toxinology 76, 63–76 (2013).
498
11. A. H. Laustsen, How can monoclonal antibodies be harnessed against neglected tropical
499
diseases and other infectious diseases?, Expert Opin. Drug Discov. 14, 1103–1112 (2019).
500
12. R. A. Harrison, J. M. Gutiérrez, Priority Actions and Progress to Substantially and
501
Sustainably Reduce the Mortality, Morbidity and Socioeconomic Burden of Tropical Snakebite,
502
Toxins 8 (2016).
503
13. A. H. Laustsen, M. Engmark, C. Clouser, S. Timberlake, F. Vigneault, J. M. Gutiérrez, B.
504
Lomonte, Exploration of immunoglobulin transcriptomes from mice immunized with three-
505
finger toxins and phospholipases A2 from the Central American coral snake, Micrurus
506
nigrocinctus, PeerJ 5, e2924 (2017).
507
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
29
14. N. H. Tan, K. Y. Wong, C. H. Tan, Venomics of Naja sputatrix, the Javan spitting cobra: A
508
short neurotoxin-driven venom needing improved antivenom neutralization, J. Proteomics 157,
509
18–32 (2017).
510
15. A. H. Laustsen, J. M. Gutiérrez, A. R. Rasmussen, M. Engmark, P. Gravlund, K. L. Sanders,
511
B. Lohse, B. Lomonte, Danger in the reef: proteome, toxicity, and neutralization of the venom of
512
the olive sea snake, Aipysurus laevis, Toxicon 107, 187–196 (2015).
513
16. C. H. Tan, K. Y. Tan, S. E. Lim, N. H. Tan, Venomics of the beaked sea snake, Hydrophis
514
schistosus: A minimalist toxin arsenal and its cross-neutralization by heterologous antivenoms, J.
515
Proteomics 126, 121–130 (2015).
516
17. A. H. Laustsen, M. Engmark, C. Milbo, J. Johannesen, B. Lomonte, J. M. Gutiérrez, B.
517
Lohse, From Fangs to Pharmacology: The Future of Snakebite Envenoming Therapy, Curr.
518
Pharm. Des. 22, 5270–5293 (2016).
519
18. R. M. Kini, S. S. Sidhu, A. H. Laustsen, Biosynthetic Oligoclonal Antivenom (BOA) for
520
Snakebite and Next-Generation Treatments for Snakebite Victims, Toxins 10, 534 (2018).
521
19. A. H. Laustsen, A. Karatt-Vellatt, E. W. Masters, A. S. Arias, U. Pus, C. Knudsen, S. Oscoz,
522
P. Slavny, D. T. Griffiths, A. M. Luther, R. A. Leah, M. Lindholm, B. Lomonte, J. M. Gutiérrez,
523
J. McCafferty, In vivo neutralization of dendrotoxin-mediated neurotoxicity of black mamba
524
venom by oligoclonal human IgG antibodies, Nat. Commun. 9, 3928 (2018).
525
20. A. H. Laustsen, Snakebites: costing recombinant antivenoms, Nature 538, 41–41 (2016).
526
21. A. H. Laustsen, K. H. Johansen, M. Engmark, M. R. Andersen, Recombinant snakebite
527
antivenoms: A cost-competitive solution to a neglected tropical disease?, PLoS Negl. Trop. Dis.
528
11, e0005361 (2017).
529
22. T. P. Jenkins, A. H. Laustsen, Cost of Manufacturing for Recombinant Snakebite
530
Antivenoms, Front. Bioeng. Biotechnol. 8 (2020).
531
23. D. J. Schofield, A. R. Pope, V. Clementel, J. Buckell, S. D. Chapple, K. F. Clarke, J. S.
532
Conquer, A. M. Crofts, S. R. Crowther, M. R. Dyson, G. Flack, G. J. Griffin, Y. Hooks, W. J.
533
Howat, A. Kolb-Kokocinski, S. Kunze, C. D. Martin, G. L. Maslen, J. N. Mitchell, M.
534
O’Sullivan, R. L. Perera, W. Roake, S. P. Shadbolt, K. J. Vincent, A. Warford, W. E. Wilson, J.
535
Xie, J. L. Young, J. McCafferty, Application of phage display to high throughput antibody
536
generation and characterization, Genome Biol. 8, R254 (2007).
537
24. C. D. Martin, G. Rojas, J. N. Mitchell, K. J. Vincent, J. Wu, J. McCafferty, D. J. Schofield, A
538
simple vector system to improve performance and utilisation of recombinant antibodies, BMC
539
Biotechnol. 6, 46 (2006).
540
25. J. D. Marks, A. D. Griffiths, M. Malmqvist, T. P. Clackson, J. M. Bye, G. Winter, By–
541
Passing Immunization: Building High Affinity Human Antibodies by Chain Shuffling,
542
Bio/Technology 10, 779–783 (1992).
543
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
30
26. G. Richard, A. J. Meyers, M. D. McLean, M. Arbabi-Ghahroudi, R. MacKenzie, J. C. Hall,
544
In Vivo Neutralization of α-Cobratoxin with High-Affinity Llama Single-Domain Antibodies
545
(VHHs) and a VHH-Fc Antibody, PLOS ONE 8, e69495 (2013).
546
27. A. H. Laustsen, V. Greiff, A. Karatt-Vellatt, S. Muyldermans, T. P. Jenkins, Animal
547
Immunization, in Vitro Display Technologies, and Machine Learning for Antibody Discovery,
548
Trends Biotechnol. (2021), doi:10.1016/j.tibtech.2021.03.003.
549
28. N. R. Casewell, T. N. W. Jackson, A. H. Laustsen, K. Sunagar, Causes and Consequences of
550
Snake Venom Variation, Trends Pharmacol. Sci. 41, 570–581 (2020).
551
29. M. Lewin, S. Samuel, J. Merkel, P. Bickler, Varespladib (LY315920) Appears to Be a
552
Potent, Broad-Spectrum, Inhibitor of Snake Venom Phospholipase A2 and a Possible Pre-
553
Referral Treatment for Envenomation, Toxins 8, 248 (2016).
554
30. H. S. Rasmussen, P. P. McCann, Matrix Metalloproteinase Inhibition as a Novel Anticancer
555
Strategy: A Review with Special Focus on Batimastat and Marimastat, Pharmacol. Ther. 75, 69–
556
75 (1997).
557
31. C. L. Dobson, P. W. A. Devine, J. J. Phillips, D. R. Higazi, C. Lloyd, B. Popovic, J. Arnold,
558
A. Buchanan, A. Lewis, J. Goodman, C. F. van der Walle, P. Thornton, L. Vinall, D. Lowne, A.
559
Aagaard, L.-L. Olsson, A. Ridderstad Wollberg, F. Welsh, T. K. Karamanos, C. L. Pashley, M.
560
G. Iadanza, N. A. Ranson, A. E. Ashcroft, A. D. Kippen, T. J. Vaughan, S. E. Radford, D. C.
561
Lowe, Engineering the surface properties of a human monoclonal antibody prevents self-
562
association and rapid clearance in vivo, Sci. Rep. 6, 38644 (2016).
563
32. T. Jain, T. Sun, S. Durand, A. Hall, N. R. Houston, J. H. Nett, B. Sharkey, B. Bobrowicz, I.
564
Caffry, Y. Yu, Y. Cao, H. Lynaugh, M. Brown, H. Baruah, L. T. Gray, E. M. Krauland, Y. Xu,
565
M. Vásquez, K. D. Wittrup, Biophysical properties of the clinical-stage antibody landscape,
566
Proc. Natl. Acad. Sci. 114, 944–949 (2017).
567
33. M. R. Dyson, E. Masters, D. Pazeraitis, R. L. Perera, J. L. Syrjanen, S. Surade, N.
568
Thorsteinson, K. Parthiban, P. C. Jones, M. Sattar, G. Wozniak-Knopp, F. Rueker, R. Leah, J.
569
McCafferty, Beyond affinity: selection of antibody variants with optimal biophysical properties
570
and reduced immunogenicity from mammalian display libraries, mAbs 12, 1829335 (2020).
571
34. M. Bailly, C. Mieczkowski, V. Juan, E. Metwally, D. Tomazela, J. Baker, M. Uchida, E.
572
Kofman, F. Raoufi, S. Motlagh, Y. Yu, J. Park, S. Raghava, J. Welsh, M. Rauscher, G.
573
Raghunathan, M. Hsieh, Y.-L. Chen, H. T. Nguyen, N. Nguyen, D. Cipriano, L. Fayadat-Dilman,
574
Predicting Antibody Developability Profiles Through Early Stage Discovery Screening, mAbs
575
12, 1743053 (2020).
576
35. K. Ratanabanangkoon, P. Simsiriwong, K. Pruksaphon, K. Y. Tan, B. Chantrathonkul, S.
577
Eursakun, C. H. Tan, An in vitro potency assay using nicotinic acetylcholine receptor binding
578
works well with antivenoms against Bungarus candidus and Naja naja, Sci. Rep. 8, 9716 (2018).
579
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
31
36. S.-X. Li, S. Huang, N. Bren, K. Noridomi, C. D. Dellisanti, S. M. Sine, L. Chen, Ligand-
580
binding domain of an α7-nicotinic receptor chimera and its complex with agonist, Nat. Neurosci.
581
14, 1253–1259 (2011).
582
583
584
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
32
585
Acknowledgments: LL is thankful to Georgia Bullen for general guidance in the laboratory as
586
well as to Yessica Wouters for help with data analysis. Birte Svensson is thanked for
587
discussions on SPR analysis. Figure 3 A was created using BioRender.com.
588
Funding:
589
Villum Foundation grant 00025302 (AHL)
590
The European Research Council (ERC) under the European Union's Horizon 2020
591
research and innovation programme grant no. 850974 (AHL)
592
The Novo Nordisk Foundation (NNF16OC0019248) (AHL)
593
The Hørslev Foundation (203866) (AHL)
594
Olsens Mindefold (LL)
595
Marie og M.B. Richters Fond (LL)
596
Niels Bohr Fondet (LL)
597
Torben og Alice Fritmodts Fond (LL)
598
William Demant Fonden (LL)
599
Otto Mønsteds Fond (LL)
600
Knud Højgaards Fond (LL)
601
Rudolph Als Fondet (LL)
602
Augustinus Fonden (LL)
603
Tranes Fond (LL)
604
Author contributions:
605
Conceptualization: AHL, LL, AKV
606
Methodology: AHL, LL, AKV, JMC, PS, KB, JMG, BL
607
Investigation: AHL, LL, UP, JW, PV, KB, EWM, ASA, SO, DTG, AML, ML, RAL, HA,
608
JMC, JMG, BL, MSM
609
Visualization: LL, KB
610
Writing – original draft: LL, AHL, KB
611
Writing – review & editing: LL, AHL, AKV, JMG, BL, PS
612
Competing interests: Authors declare that they have no competing interests.
613
614
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
.CC-BY-ND 4.0 International licenseavailable under a
(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
The copyright holder for this preprintthis version posted September 7, 2021. ; https://doi.org/10.1101/2021.09.07.459075doi: bioRxiv preprint
