SCIENCE science.org 10 DECEMBER 2021 • VOL 374 ISSUE 6573 1337-C
RESEARCHmigration, which relate the rates
of both processes to fundamen-
tal interaction energies in metal
nanoparticle-support combina-
tions. Using kinetic simulations
for hundreds of such pairs, the
authors show a universal volcano
dependence of the sintering
kinetics on the metal-support
binding energy that can serve
as a single descriptor to predict
nanoparticle growth rates. The
revealed scaling relations are a
good start in the development of
high-throughput screening com-
putational approaches to drive
discovery of sintering-resistant
nanocatalysts. —YS
Science, abl9828, this issue p. 1360
VACCINES
Boosting adjuvant activity
Adjuvants can enhance immune
responses against poorly immu-
nogenic antigens, making them
an important component of
many vaccines despite the lim-
ited number currently approved
for clinical use. Silva et al. devel-
oped a new adjuvant formed by
the Toll-like receptor 4 agonist
monophosphoryl lipid A (MPLA)
self-assembled into nanoparticle
structures with detergent-like
saponin. Compared with a panel
of tested adjuvants, includ-
ing other saponin-based
formulations, saponin–MPLA
nanoparticles (SMNPs) induced
superior antibody and ger-
minal center responses after
immunization with a poorly
immunogenic HIV antigen in
both mice and rhesus macaques.
SMNPs increased lymphatic
flow in a mast cell–dependent
manner and increased anti-
gen delivery to draining lymph
nodes, suggesting mechanisms
by which saponin adjuvants act
to enhance humoral immune
responses. —CO
Sci. Immunol. 6 , eabf1152 (2021).
CANCER
Growing tumor blood
vess e l s, STAT
The co-transcriptional activa-
tors YAP/TAZ induce changes
in gene expression in endothe-
lial cells that promote tumor
angiogenesis. Shen et al. sought
to understand how YAP/TAZ are
activated in colorectal cancers
and melanomas. Cytokines
abundant in the tumor micro-
environment promoted the
interaction of YAP/TAZ with
the transcription factor STAT3.
This interaction enabled the
nuclear translocation of YAP/
TAZ. Pharmacological inhibition
or genetic ablation of YAP/TAZ
or STAT3 in endothelial cells
reduced the growth of colorectal
cancers and melanomas in mice.
—WW
Sci. Signal. 14 , eabj8393 (2021).PHASE-CHANGE MEMORY
Single element switch
Phase-change materials
are attractive for computer
memory and switching, in part
due to their small size and fast
switching speeds. However,
competitive materials frequently
have many elements, which
decreases the switching reli-
ability. Shen et al. built a pure
tellurium device that is capable
of fast switching through a
phase transformation (see the
Perspective by Calarco and
Arciprete). Unlike many other
phase-change materials, the
change in resistance happens
because the tellurium melts dur-
ing the switching process. The
resulting device can be switched
100 million times before failure
and is an appealing route
for avoiding the issues from
multi-element phase-change
materials. —BG
Science, abi6332, this issue p. 1390;
see also abm7316, p. 1321CORONAVIRUS
A look at variant-specific
boosters
The evolution of severe acute
respiratory syndrome corona-
virus 2 (SARS-CoV-2) variants
of concern (VOCs) has raised
the question of whether current
COVID-19 vaccines protect
against VOCs and if a variant
specific vaccine may be needed.
Of the currently identified VOCs,
the Delta variant is believed to be
the most transmissible, whereasthe Beta variant appears to
be the most vaccine resistant.
Corbett et al. looked at the effect
of vaccine boosting using either
the original WA-1 strain vac-
cine or a Beta variant–specific
booster. Around 6 months after
the primary two-dose vaccine
series, a third boost vaccination
resulted in higher neutral-
izing antibody levels against
VOCs in nonhuman primates.
Regardless of whether the boost
was from the original vaccine
or the Beta-specific version,
similar increases in neutralizing
antibody levels were observed
and resulted in enhanced viral
protection. —PNK
Science, abl8912, this issue p. 1343QUANTUM CHEMISTRY
Improving DFT with
deep learning
In the past 30 years, density
functional theory (DFT) has
emerged as the most widely
used electronic structure
method to predict the properties
of various systems in chemistry,
biology, and materials science.
Despite a long history of suc-
cesses, state-of-the-art DFT
functionals have crucial limita-
tions. In particular, significant
systematic errors are observed
for charge densities involv-
ing mobile charges and spins.
Kirkpatrick et al. developed a
framework to train a deep neural
network on accurate chemical
data and fractional electron con-
straints (see the Perspective by
Perdew). The resulting functional
outperforms traditional function-
als on thorough benchmarks for
main-group atoms and mol-
ecules. The present work offers
a solution to a long-standing
critical problem in DFT and
demonstrates the success of
combining DFT with the modern
machine-learning methodol-
ogy. —YS
Science, abj6511, this issue p. 1385;
see also abm2455, p. 1322