Corrections or additions?
This article by Michele Alperin was prepared for the February 14,
2001 edition of U.S. 1 Newspaper. All rights reserved.
Physiome: Simulations of Human Cells
Biochemists, molecular biologists, and medical
researchers
in laboratories across the world investigate the complex and even
noble phenomena of human biology. Each of the structures, processes,
and interactions they study yields a tiny piece of a puzzle so
enormous
that it falls beyond the grasp of human understanding. At the same
time, genomic research is revealing thousands of new proteins to
investigate,
and the old ways of science are not up to the task.
Whereas laboratory experiments alone made sense when there were only
a few proteins and genes to explore, "now scientists are dealing
with thousands, and they must simulate. They can’t do it any other
way," says Jeremy Levin, president and CEO of Physiome Sciences,
a 40-person firm at 307 College Road. What’s happening, says Levin,
is "a complete revolution in the way scientists are thinking."
Currently pharmaceuticals spend hundreds of millions of dollars
testing
potential drug compounds in the laboratory, and they often fail just
as they are ready to become products. "This," says Levin,
"is like building an oil refinery and hoping it will work in the
end." To avoid wasting vast sums of money, he explains, scientists
must do dry bench experiments, that is, virtual experiments on the
computer, before working in the laboratory at "wet bench"
experiments.
Levin has steered Physiome in that direction. To bring simulation
to the foot soldiers of biochemical research, Physiome is perfecting
a vehicle to gather experimental data and produce visual models of
human proteins, cells, tissues, and organs. The result: researchers
can work with a "perpetual cell," a simulated computer model
that can be changed and modified. On February 5 it announced a
licensing
agreement with Janssen Pharmaceutica (the Titusville-based subsidiary
of Johnson & Johnson), and it will triple its space with a move later
this year to 25,000 square feet at 150 College Road.
Levin’s focus on cell software represents a dramatic change. Since
its founding in 1993, Physiome Sciences had been identified with the
throbbing, 3-D human heart displayed on screens by its supercomputers
(U.S. 1, May 13, 1998). But when Levin came on board, he found that
what the pharmaceutical companies wanted was — not finished organs
and disease modules — but the capability to handle the volume
of gene and protein data, and to simulate them.
Levin decided not to deliver "finished organs" but to provide
a simple, standardized modeling platform. This would help the
pharmaceutical
companies develop the internal capability to understand their own
data and incorporate it in models.
He based his tactical decision on the three critical challenges facing
pharmaceutical companies today.
protein data that is about to multiply geometrically;
in the body; and
ensuring standardized experimental methods.
environment — is reported to be easy to use, flexible, and able
to communicate data easily. Whereas modeling has been restricted to
trained modelers who have created their models in standard computer
languages, Physiome’s ISC supports uniform formatting of modeling
data that can easily be shared over the Internet.
No longer are scientific colleagues left with the impossible task
of reconstructing models from information in scientific publications.
"By giving academics and those who are developing models a
centralized
capability to bank and withdraw models," says Levin, "we
provide
a critical resource for the whole field of modeling." Because
ISC standardizes the model-building process, it can be used by a
variety
of departments, from cardiovascular to oncology.
ISC also supports evolving models that seamlessly absorb legacy data,
accumulated from previous experiments, and incorporate genomic data.
The quantity of data is daunting; 50,000 to 100,000 genes will
translate
to hundreds of thousands of proteins in a cell — all of them not
yet discovered. "What is important is that cell types can adopt
and absorb the data that is emerging from the Genome Project and
become
more enriched over time," says Levin.
Physiome does not try to reproduce every aspect of every cell. "In
the same way that biology has created a natural template for a cell,
there are certain key genes that you need in these cells,"
explains
Levin. "All of our cells have those genes." Physiome supplies
its client partners with template cells that can be adjusted, based
on the scientific intentions and needs of the user, to correspond
to a particular cell type. For example, the user of ISC can mutate
the cell by changing it from male to female or from a kidney cell
to a heart or a brain cell, through a sophisticated process called
computational cloning.
The goal of the ISC modeling environment is to enable both Physiome’s
own researchers, and as many different scientific constituencies as
possible, to develop, store, and exchange biological models. "The
whole idea is to create a common language so that anyone working in
modeling biology on the computer can develop models that speak the
same language," says Thomas Colatsky, executive vice president
and chief scientific officer of Physiome.
Speaking with fervor about the "living" creations of ISC,
Levin says, "Now instead of having a living cell that eventually
dies, you have a perpetual cell within the computer that you can
experiment
with to your heart’s delight. It lives forever."
Modeling virtual cells saves enormous amounts of time. To create a
new living cell type in the laboratory and make it stable requires
four to six months. The time it would take to do this for the 100,000
cells out there, says Levin, would be far more than the lifetimes
of all the scientists in this country.
But these models are more than time savers. They produce results that
human beings might never be able to predict. Referring to the cellular
proteins and interactions modeled by ISC, Colatsky explains, "They
all add up, in a nonintuitive way, to output which is the behavior
of a cell. It is hard to predict what will happen. It is not a linear
process."
Translating Levin’s vision into "virtual" reality required
substantial capital. In May, 1999, his effort to raise $25-30 million
was vastly oversubscribed. He got $100 million in commitments, but
Physiome’s board decided that $50 million was sufficient to achieve
its business plan. With money in hand, Physiome battened down the
hatches. Explains Levin: "We buried ourselves, with no intent
to talk to our pharmaceutical partners, and focused on developing
a superb technological base."
Now they have come back to the living, with software that can
reproduce
biological form and function. With pride, Levin comments, "We
are there. InSilico Cell is a beautiful platform." His licensing
deal with Janssen uses one of InSilico Cell’s direct applications,
CardioPrism, which simulates cardiac cell behavior and evaluates the
suitability of proteins as targets.
In his 47 years, Jeremy Levin has lived in many worlds,
both geographically and intellectually. He was born in South Africa,
where his father was a direct contemporary of Nelson Mandela. In 1960,
when Levin was about seven, his family fled to Rhodesia. The apartheid
government had asked his father to choose between jail or working
for his political opponents. Levin remembers, "We were so poor
we had to drive across the border to Rhodesia."
In Rhodesia, his father founded the first newspaper for blacks, and
the family quickly became isolated. Just prior to the Ian Smith’s
unilateral declaration of independence, the white government decided
to arrest Levin’s father. He was able to flee when some World War
II army buddies in the Rhodesian security forces warned him of the
impending arrest. Levin’s father left with his sister, and he was
left behind with his mother and brother.
Thinking about his father, who sacrificed such a tremendous amount
for his principles, Levin speaks with admiration, "My father
believed
in truth, justice and creativity. He also believed that what mattered
most was making a difference in whatever you do, but that ultimately
making your society a better place to live was what mattered more
than anything else."
Levin finally made it to England, where he attended a military
boarding
school — all his family could afford. For high school he enrolled
in an early experimental school, whose population was "very
working
class." Being a new immigrant made things difficult. "I was
a foreigner, with a funny accent," he says. Assessing his overall
high school experience from decades beyond, Levin says, "I was
a great rugby player in a useless high school. I got 2 Cs and 2 Ds
at the end of high school, but that was the best in the school."
But with the help of an English teacher at the school, he eventually
made it to Oxford University, on scholarship, as part of an experiment
to see whether graduates from a school like his could succeed at the
university. In Levin’s case, the experiment worked, leading to a
success
story that Levin ruefully calls "kid makes good from foreign
land."
At Oxford, he began an amazing academic career, with top grades,
awards,
and scholarships. He received a scholarship to do a PhD in DNA
structure
at Oxford and a simultaneous scholarship to study pre-clinical
medicine.
In 1978, he moved to Cambridge for clinical medicine, again with a
scholarship, and finished his MD degree in 1981. Levin practiced
medicine
in a number of hospitals and, in 1984, became a European Fellow at
the University Hospital in Geneva, where — while skiing —
he made a contact that changed his life. He met an investment banker
who was evaluating a new biotechnology company. The company was
developing
drugs projected to come to market in two years, which surprised Levin,
who knew from his Oxford work that this estimate was not realistic.
What impressed Levin deeply was the realization that this investment
banker was willing to invest $50 million in a creative biotech
company,
despite estimates that were "out of whack." Says Levin, "I
was totally enthused by the way the investment community would back
the innovative science that the company was showing me."
So began Levin’s life-long love affair with biotechnology. Although
Levin appreciated the day-to-day interactions of medical practice,
he felt stymied by its limitations. "Medicine doesn’t help the
great mass of people," says Levin, echoing his father’s devotion
to improving human society. "Biotechnology offered the possibility
to help an enormous number of people. We couldn’t hope to treat AIDS
and an avalanche of other infectious diseases without it," he
says. "That’s why I find innovative technologies fascinating."
After his transformative meeting with the investment banker, Levin
decided he was ready to be recruited by an American firm. His first
business position, in 1985, was as medical director of Focus
Technologies
in Washington, D.C. The driving idea at Focus, says Levin, "was
that you could predict a disease, find its genetic marker, and then
prevent the disease." While at Focus, Levin worked with a
strategic
consulting firm, Corporate Decisions, and identified an opportunity
in genetic testing. In order to pursue this opportunity, Levin decided
to leave Focus.
Levin raised a tiny amount of money from Crest Ventures in San
Francisco,
bought out a genetic testing company from a public company, turned
the company around, and, in 1990, sold it to Genzyme and "did
well." Levin worked for two years with Genzyme as a vice president
doing mergers and consolidations, and in 1992 was recruited to be
CEO of Cadus Pharmaceuticals in Tarrytown, New York.
The following year, in 1993, Levin was asked by Oxford professor Denis
Noble to look at Noble’s work of 30 years, single-cell models of
cardiac
tissues. Levin says, "I thought that the technology he showed
me, if properly developed, heralded a complete change in drug
discovery.
If properly commercialized, it would be wonderful." Levin promptly
went to friends and family and reached into his own pockets and raised
$2.5 million to continue the research. (Other than talking about his
father and his childhood, Levin declines to discuss his current family
life.) In 1997, as chairman of Physiome, he hired Bill Scott to begin
assembling the infrastructure that has grown into the team at Physiome
today.
Meanwhile, during Levin’s five years at Cadus, he
collaborated
with Bristol-Myers Squibb and other pharmaceutical companies and took
the company public. At the end of 1998 he left Cadus to found the
Perseus Soros Fund, which invests in small biotech companies with
promising products and technologies.
Levin now had to prove that Noble’s technology would work in the down
and dirty world of scientific experimentation. And he did. Physiome
was able to simulate the effect of a particular drug on its virtual
models of the human heart, and to show, to the FDA’s satisfaction,
that it could predict the consequences of using that drug. An apparent
heart problem caused by the drug was proved to be not a potentially
damaging side effect.
In late summer of 1999, Scott indicated that he wanted to step back
from running a company and just sit on boards. Physiome’s board began
looking for a CEO, and despite other outstanding candidates, they
came to Levin to give him one last chance to take the job. Levin
became
Physiome’s CEO last year.
"I love operating companies," says Levin. "I’ve done
mergers
and acquisitions, lots of financing, and a tremendous amount of
building
companies from the ground up — from washing dishes to being
chairman.
I’m not afraid to do anything." He says his philosophy is: "If
you haven’t done it, don’t ask someone else to do it," and if
you cannot do it yourself, then you should hire really smart people
to do it for you.
Physiome’s products support pharmaceutical companies primarily in
the pre-clinical drug discovery area, with some overlap into phase
one, the clinical trials to test safety.
companies decide, early on, whether a product is likely to be
effective
in human beings and, if so, in what types of human beings. For
example,
they can insert genetic variants in Physiome’s model and then
construct
trials to exclude or include variants. Or, researchers can construct
models of male and female cells and assess how a drug might affect
men and women differently. They can use the models to identify drugs
likely to have a toxic effect or to choose which of several dozen
compounds should be put in the testing pipeline.
the genes that are critical in human diseases is one challenge. And,
says Levin, because genes work in complex inter-relationships,
evaluating
one gene at a time can, and does, give spurious results. The
consequence
can be very expensive mistakes, as scientists move out of the test
tube and into animal and human testing.
Five years ago the idea of a computer experiment before
a bench experiment was ludicrous. Now, an array of software
applications
tools are supported by Physiome’s modeling platform, ISC, which has
its roots in the 3-D virtual heart, the organ model for which the
company achieved fame. This ISC environment can be compared, says
Levin, to a Microsoft Office where the work of the day is biological
modeling rather than the creation of budgets, reports, and business
presentations.
Just as Microsoft Office supports numerous applications that support
specific office functions, so does InSilico Cell support modeling
applications that focus on specific diseases or therapies. The
platform
also offers specialized tools that enable the modeling of a range
of diseases and molecules. A number of the tools and applications
are web-enabled:
1. CardioPrism, a suite of cardiac models, helps assess and
predicts drug-induced heart abnormalities — how a chemical might
cause an abnormality of a heartbeat or how it could interrupt an
abnormality.
CardioPrism also assesses whether a compound will be toxic, from the
cell through the organ levels.
A pharmaceutical firm can license a model framework of CardioPrism
and customize it using Physiome’s own rich databases. Or it can create
models behind a firewall. "They are able to take a look at
something
unique without having someone else look at the same data," says
Levin. The latest licensing of this application was to Janssen, and
he sees a huge marketplace for it.
But the heart is no longer the entire focus. "Physiome started
with a whole heart," says Levin, "but now we are capable of
and have started building lungs, bladder, and other major products
important to the pharmaceutical industry."
Physiome is also developing models of immune system cells, which will
be ready shortly, and framework models of blood coagulation.
overlay drugs that have run into trouble at the FDA. By testing a
drug on the Browser, researchers can see how each drug causes its
side effects. The goal is to help customers and others better
understand
what classes of drugs cause abnormalities in the heart. The issue
of side effects on the heart is an enormously important problem: three
out of seven drugs that fail the FDA are rejected because of
abnormalities
they caused in the heart. Every major class of drugs has some effect
on the heart — even some antibiotics and antihistamines.
organize, and evaluate the incredibly complicated protein pathways
that signal events on a cell surface and transmit that signal into
the interior of the cell. Abnormalities in these pathways form the
basis of diseases as varied as inflammation, cancer, asthma,
coagulation
cascades, and cardiac abnormalities. The faulty signals in these
pathways
are important drug targets.
For example, a series of abnormal pathways occur in cancer. Each
pathway
may comprise as many as 50 different proteins that interact with each
other in different configurations, yielding a huge number of potential
interactions. Only a small number will be faulty. Whereas a human
being can analyze only a few interactions, the Pathway Editor
evaluates
all possible faulty signals and highlights the most likely drug
targets
for further consideration.
build whole cells.
Physiome partners with academic researchers and
institutions
across the country and the world for its beta testing. Princeton
University’s
Jim Broach, for instance, is studying the response of yeast cells
to changes in the environment.
Using the virtual model he creates with ISC, Broach can stimulate
the virtual cell with a particular nutrient and watch the effects
on the biochemical pathway controlling the growth of the yeast cell.
If the results in the virtual cell do not match what he sees in the
living cell, he knows the model is missing something. If they do
match,
however, he cannot conclude that the model correctly represents the
cell, but only that it is consistent with the cell’s actual behavior.
"We need software like this to help us understand complex
biological
systems," says Broach. "There is a continuing convergence
of biology, physics, and computer science that will define how we
do biology in the next millennium or century. Physiome is on the
forefront
of taking that step."
And what, in Broach’s view, does Physiome stand to gain from sharing
its software? "Physiome is learning how to refine their models
and their software on the basis of the successes and failures we have
in modeling the biology of this relatively simple microorganism,"
says Broach. "That gives them the ability to use the software
more effectively with more complicated cells, and then they can
interact
with large pharmaceuticals or biotechs to pursue drug discovery."
Physiome’s agreements with laboratories like Broach’s give Physiome
exclusive rights to the technology that emerges. Physiome does not
"own" the scientists, says Levin, and, on the contrary, wants
them to publish as much and as frequently as possible. With groups
at the University of Aukland and Oxford University, however, Physiome
also has rights to research and inventions achieved with the software.
Physiome is actively working to make CellML the world-wide standard
for biological modeling. "The goal of the CellML website,"
maintains Levin, "is to make it possible for as many different
constituencies as possible to develop and create biological
models."
To support this effort, Physiome is attempting to create a single
set of standards for biological modeling. "Corporate goals are
served by having everyone playing on a level playing field with regard
to modeling," says Levin. Levin likens this standardization effort
to the early struggles between Beta Max and VHS, where the adoption
of a single standard helped everyone to produce better products and
to get wider acceptance for them.
An alternative view is offered by Christina Kalb, a spokeperson for
Entelos, a company that has models for simulating human disease.
Entelos
also has a standardized method and language for developing its models
but does not want to make them publicly available as tools. "We
think it is a little early for standardization," she says.
"Usually
you don’t know what your language needs to be capable of until you’ve
encountered a lot of different problems." Until researchers have
identified all of the problems they want to address with the
technology,
Kalb believes that standardization may limit rather than enhance.
Physiome, however, is promoting CellML as the standard and wants to
preempt the creation of other modeling languages. "Our goal is
to have CellML used by all," says Levin. To promote CellML,
Physiome
has advertised widely and created an impressive advisory board to
develop modeling standards. The board comprises scientists from
diverse
areas of research, to ensure representation of a broad spectrum of
biochemistry. "We want to make sure that the specifications for
the language actually incorporate as much real biology as they
can,"
explains Colatsky. "We don’t want to take a limited view of what
models have to do."
At the board’s first teleconference, which included scientists from
New Zealand, Great Britain, and across the United States, members
agreed on a charter, goals, and near-term milestones. Colatsky sees
that first teleconference as revolutionary, "because no one has
been able to bring together as much diversity in biology and try to
unify it with a common approach." The advisory board will meet
face to face in March.
Peter Kramer, a business development consultant in the biotech
industry,
injects a note of caution. "Today, we just don’t have a sufficient
understanding of biological processes or information on protein
pathways
to enable simulations to predict useful results without extensive
biological experimentation," says Kramer. Kramer does see
Physiome’s
modeling capabilities as a potential bridge to a future in which
models
are more reliably predictive.
Levin realizes that the company’s success will hinge on a number of
variables — advances in technology, ability to recruit top-level
people, availability of capital, and a healthy pharmaceutical
industry.
Yet he maintains that science and business are about to be partners
in the modeling enterprise.
"In the past year," he says, "modeling has exploded in
the academic world, and it is about to explode in the business
world."
Physiome plans to provide the pharmaceutical companies with modeling
capabilities that can accommodate the huge amounts of data on the
horizon. By enabling pharmaceutical companies to do modeling for
themselves,
using their own data, says Levin, "We are providing a new
industrial
base for pharmaceutical companies."
Musing on how his own path is similar to that of his father, Levin
says, "I have tried to follow the same guiding principles."
Embracing his father’s belief in using creativity to make the world
a better place, Levin believes that Physiome’s technology will
eventually
change the pharmaceutical industry and "will provide a completely
new way of looking at our living world."
"Physiome is a completely different type of company," says
Levin. "We are breaking down the walls between computers and
biology
in extraordinarily creative ways." And meditating on his own role,
Levin says, "What an honor to have the chance to participate in
that process."
08540. Jeremy M. Levin, CEO. 609-987-1199; fax, 609-987-9393.
Corrections or additions?
This page is published by PrincetonInfo.com
— the web site for U.S. 1 Newspaper in Princeton, New Jersey.
Facebook Comments