810 22 MAY 2020 • VOL 368 ISSUE 6493 sciencemag.org SCIENCE
IMAGE: NIAID/CC BY/FLICKRCrotty and Alessandro Sette, immuno-
logists at the La Jolla Institute for
Immunology, used bioinformatics tools to
predict which segments of the virus’ pro-
teins should stimulate T cells most effec-
tively. They then exposed immune cells from
10 patients who had recovered from mild
cases of COVID-19 to these viral snippets.
All of the patients carried helper T cells
that recognized the SARS-CoV-2 spike pro-
tein, which enables the virus to infiltrate
our cells. They also harbored helper T cells
that reacted to other SARS-CoV-2 proteins.
And in 70% of the subjects, the team de-
tected virus-specific killer T cells, they
reported last week in Cell. “The immune
system sees this virus and mounts an effec-
tive immune response,” Sette says.
The results jibe with those of a study
posted as a preprint on medRxiv on
22 April by immunologist Andreas Thiel of
the Charité University Hospital in Berlin
and colleagues. They identified helper T
cells targeting the spike protein in 15 out
of 18 patients hospitalized with COVID-19.
Before these studies, researchers didn’t
know whether T cells played a role in
eliminating SARS-CoV-2—or even whether
they could provoke a dangerous immune
system overreaction. “These papers are
really helpful because they start to define
the T cell component of the immune re-
sponse,” Rasmussen says.
She and other scientists caution that
the results do not mean people who have
recovered from COVID-19 are safe from re-
infection. But they do raise hopes that a vac-
cine could give lasting protection against
the virus. To spark production of antibodies,
a vaccine needs to stimulate helper T cells,
Crotty notes. “It is encouraging that we are
seeing good helper T cell responses against
SARS-CoV-2 in COVID-19 cases,” he says.
The results have other significant impli-
cations for vaccine design, says molecular
virologist Rachel Graham of the University
of North Carolina, Chapel Hill. Most vac-
cines under development aim to elicit an
immune response against the spike protein,
but the La Jolla group’s study determined
that T cells reacted to several viral proteins,
suggesting vaccines that incite an immune
response to these proteins as well could be
more effective. “It is important to not just
concentrate on one protein,” Graham says.
Both teams also wondered whether
people who haven’t been infected with
SARS-CoV-2 also produce T cells that could
combat it. Thiel and colleagues analyzed
blood from 68 uninfected people and
found that 34% hosted helper T cells that
recognized SARS-CoV-2. The La Jolla team
studied stored blood samples collected be-
tween 2015 and 2018, well before the cur-
rent pandemic began, and detected these
cross-reactive helper T cells in about half
of them. The researchers think these cells
were likely triggered by past infection with
one of the four human coronaviruses that
cause colds; proteins in these viruses re-
semble those of SARS-CoV-2.
The studies don’t show people with this
cross-reactivity are less likely to become
ill from COVID-19. But viral immunologist
Steven Varga of the University of Iowa says
the results do suggest “one reason that a
large chunk of the population may be able
to deal with the virus is that we may have
some small residual immunity from our ex-
posure to common cold viruses.” jImmune hunters called T cells can seek and destroy a cell (green) infected with SARS-CoV-2 (yellow).
T
he international alarm about the
COVID-19 pandemic was sounded first
not by a human, but by a computer.
HealthMap, a website run by Boston
Children’s Hospital, uses artificial in-
telligence (AI) to scan social media,
news reports, internet search queries, and
other data for signs of disease outbreaks.
On 30 December 2019, it spotted a news re-
port of a new type of pneumonia in Wuhan,
China, and issued a one-line email bulletin
that seven people were in critical condition,
rating the urgency at three on a scale of five.
Humans weren’t far behind. Colleagues
in Taiwan had already alerted Marjorie
Pollack, a medical epidemiologist in New
York City, to social media chatter in China
that reminded her of the 2003 outbreak of
severe acute respiratory syndrome (SARS),
which spread to dozens of countries and
killed 774. “It fit all of the been there, done
that déjà vu for SARS,” Pollack says. Less
than 1 hour after the HealthMap alert, she
posted a more detailed notice to ProMED ,
a list server with 85,000 subscribers for
which she is a deputy editor.
But the early alarm from HealthMap
underscores the potential of AI, or ma-
chine learning, to keep watch for conta-
gion. As the COVID-19 pandemic continues
to spread, AI researchers are teaming with
tech companies to build automated track-
ing systems that will mine social media,
news reports, and public health data for
signs of new outbreaks. AI is no substitute
for traditional public health monitoring,
cautions Matthew Biggerstaff, an epidemio-
logist with the U.S. Centers for Disease Con-
trol and Prevention (CDC). “This should be
viewed as one tool in the toolbox,” he says.
But it can fill a need, says Elad Yom-Tov,
a Microsoft computer scientist who has
worked with public health officials in theBy Adrian ChoCOVID 19AI systems aim
to sniff out
coronavirus
outbreaks
Alerts would not substitute
for on-the-ground sleuthing,
experts warn
Published by AAAS