Research highlights
Drug addiction
Monoclonal
antibody to treat
fentanyl overdose
The opioid receptor antago-
nist, naloxone, is currently the
only available treatment for
reversing the negative effects
of fentanyl overdose, includ-
ing respiratory depression.
However, naloxone cannot be
administered prophylactically,
and multiple administrations
are typically necessary.
Here, Bremer et al. profile the
opioid binding activity of
the fully human monoclonal
antibody (mAb) CSX-1004,
designed to bind fentanyl. The
mAb recognized a wide range of
fentanyl analogues (FAs) with
picomolar affinity and did not
cross-react with endogenous
opioid peptides, prescription
opioids, or opioid antagonists.
In mice administered fentanyl
to induce a near maximal possible
effect, intravenous CSX-1004
markedly reversed fentanyl-
induced antinociception within
20 minutes. The mAb also
improved respiratory depression
induced by carfentanil (a potent
FA) within 10 minutes, surpassing
the magnitude of naloxone rever-
sal within 60 minutes of treat-
ment. In addition, pretreatment
of mice with CSX-1004 prevented
fentanyl antinociception. In rats,
CSX-1004 displayed a favourable
pharmacokinetic and toxico-
logical profile. In a non-human
primate model of opioid-induced
respiratory depression, CSX-1004
significantly antagonized the
effects of repeated fentanyl
challenges.
CSK-1004 infusion has
entered phase I testing and
has been granted fast track
designation by the FDA.
Sarah Crunkhorn
Original article: Bremer, P. T. et al.
Investigation of monoclonal antibody
CSX-1004 for fentanyl overdose.
Nat. Commun. https://doi.org/10.1038/
s41467-023-43126-0 (2023)
nature reviews drug discovery
Antibacterial agents
Explainable deep
learning discovers
novel antibiotic
Deep learning approaches have
demonstrated success in identify-
ing potential antibiotics from
chemical libraries, but the mod-
els used are not explainable and
do not provide chemical insight.
Now, Wong et al. develop
an explainable deep learning,
substructure-based approach
for the discovery of novel anti­
biotics. First, a set of 39,312 chem-
ically diverse compounds were
screened for growth-inhibitory
activity against S. aureus. Using
Chemprop, ensembles of graph
neural networks were trained
on the screening data to predict
whether a new compound will
inhibit bacterial growth based
on its chemical structure. This
training set of compounds was
counter-screened for cytotoxic-
ity in three human cell lines, to
develop orthogonal models that
predict cytotoxicity.
The trained graph neural
networks were applied to make
predictions of antibiotic activity
and cytotoxicity for 12,076,365
compounds. Explainable
graph algorithms identified
substructure-based ration-
ales for compounds with high
predicted antibiotic activity
and low predicted cytotoxicity.
Filtering steps resulted in a set
of 283 compounds, which were
experimentally tested. Two
compounds from the same struc-
tural class were selective against
methicillin-resistant S. aureus
(MRSA) and vancomycin-resistant
enterococci, overcame common
resistance determinants, and
were effective in both the topical
and systemic treatment of MRSA
in mouse infection models.
Sarah Crunkhorn
Original article: Wong, F. et al. Discovery
of a structural class of antibiotics with
explainable deep learning. Nature https://
doi.org/10.1038/s41586-023-06887-8 (2023)
Allergy
Eliminating IgE
reverses allergy
Immunoglobulin E (IgE) plays
an essential role in type 1
hypersensitivity reactions and
allergic disorders, representing
a direct target to combat allergy.
However, complete elimina-
tion of pathogenic IgE produc-
tion has not yet been clinically
achieved.
Here, Limnander et al.
present a strategy designed
to both eliminate existing
IgE-producing plasma cells
(PCs) and prevent the forma-
tion of new IgE-producing PCs.
They harness their previously
developed bispecific antibody
BCMAxCD3 — which binds B cell
maturation antibody (BCMA)
on PCs through the targeting
arm, and CD3 on T cells through
the effector arm — to induce
T cell-mediated killing of PCs.
This antibody is combined with
dupilumab, which blocks IL-4Rα
to inhibit IgE class switching.
Importantly, this approach
allows for reconstitution of
other immunoglobulin classes.
In a mouse house dust mite
exposure model of allergy,
transient depletion of PCs with
BCMAxCD3, combined with sus-
tained IL4Rα blockade, ablated
IgE production and blocked IgE
resurgence, preventing anaphy-
laxis upon allergen re-exposure.
Similarly, persistent elimina-
tion of IgE was obtained using
the combination treatment in
cynomolgus monkeys. Further-
more, the bispecific antibody
alone efficiently depleted
IgE-secreting bone marrow PCs
from allergic donors in vitro
and completely reduced IgE in
patients with multiple myeloma.
Sarah Crunkhorn
Original article: Limnander, A. et al.
A therapeutic strategy to target distinct
sources of IgE and durably reverse allergy.
Sci. Transl Med. 15, eadf9561 (2023)
Volume 23 | February 2024 | 103–107 | 107
Covalent drugs
High-throughput
profiling of reactive
cysteines
An effective approach to target
disease-relevant proteins is the
use of covalent inhibitors that
engage nucleophilic cysteine resi-
dues. Protein–ligand interactions,
including cysteine reactivity,
can be profiled using chemo-
proteomics techniques, such as
activity-based protein profiling
(ABPP) technologies. However,
adapting proteome-wide profil-
ing of cysteine ligandability for
high-throughput applications
remains challenging.
Here, Yang et al. present
a multiplexing-based high-
throughput platform in 96-well
plates, called tandem mass tag
(TMT)-ABPP, for proteome-wide
profiling of reactive cysteines.
TMT-ABPP enables routine inter-
rogation of 18,000 or 24,000
reactive cysteines based on only
10 or 20 μg of native proteome
input, respectively. Compared
to existing strategies, this
platform provides a 2–3 fold
improvement in cysteinome
coverage and a 5–10 fold
reduction in input material.
Applying the TMT-ABPP
platform to screen a library of 192
electrophiles in HEK293T cells,
the researchers mapped the ligan-
dability of more than 38,000 reac-
tive cysteines from 8,274 human
proteins, creating the largest
coverage of reactive cysteinome
within a single cell line.
The improved sensitivity
of the TMT-ABPP platform
revealed unknown cellular tar-
gets of five well-characterized
compounds and covalent
drugs — ibrutinib, THZ1,
ARS-1620, Sulfopin, and DMF,
using three different cell lines.
Sarah Crunkhorn
Original article: Yang, K. et al. Accelerating
multiplexed profiling of protein-ligand
interactions: High-throughput plate-based
reactive cysteine profiling with minimal input.
Cell Chem. Biol. https://doi.org/10.1016/
j.chembiol.2023.11.015 (2023)