#98 from R&D
Innovator Volume 3, Number 5
and HIV—You’ve Got to be Kidding!
Friedman is a graduate student at the department of pharmaceutical
chemistry, University of California, San Francisco.
You may have
heard about the seemingly unrelated topics that came together in
my Ph.D. work--the AIDS virus (HIV) and buckyballs,
buckminsterfulleres, those interesting carbon molecules that have
increasingly caught the attention of chemists in the past few
I doubt I would
even have started thinking about buckyballs—in relation to
HIV—at all except for a friend's chance comment during a
late-night lab bull session.
Diana Roe and I were members of research groups that were
looking for an inhibitor of the HIV protease, one which might
become the basis for an AIDS drug.
The HIV protease plays a critical role for virus
proliferation; therefore, and inhibitor may eventually protect
people from the disease. Talking
about work that had been done in the field, we agreed that some
truly interesting and strange molecules had been assayed as
potential HIV protease inhibitors.
As we mulled over
this list, Diana said, “Yeah, what are they going to try next,
We both chuckled,
I thought about it for a minute and responded, “You know, that
might actually work." I
was thinking, the protease active site—the physical target for
any inhibitor—is hydrophobic (in water it prefers to associate
with other hydrophobic surfaces instead of water), and it is
rather round. Likewise,
buckyballs are hydrophobic, and they are round too.
I went to the chalk board and drew a picture of a Pacman
(remember the voracious creature in that primordial video game?)
devouring the energizer dot, it's target on the screen.
The pacman represented the protease, and its jaws stood for
the flaps on the active site.
The dot, of course, was the buckyball. I drew two
negatively charged aspartates (a type of amino acid) which were
known to be a part of the active site, in the mouth of the pacman/protease.
Then I drew two positively charged amino groups on the
surface of the buckyball, forming salt bridges with the negatively
Over the next
couple of weeks the idea became a bit of a joke in our group;
people suggested that I publish it in the Journal
of Irreproducible Results.
I thought this juxtaposition of two unrelated fields was
pretty weird, but I also felt there were reasons that the right
fullerene derivative could be an effective inhibitor.
I'd spent many
hours staring at images of the HIV protease's active site, since
our department focused on a program of structurally based drug
design to block this essential enzyme.
I therefore knew that most of the active site was lined
with hydrophobic amino acids.
In water, hydrophobic surfaces tend to associate with each
other, much in the same way that drops of oil on the surface of
water coalesce into larger drops.
Fullerenes are also hydrophobic.
active site is large, open and tube-like (a circular cross
section), with two flaps that wrap around its natural substrates,
an arrangement that struck me as ideal for interacting with the
spherical fullerene. The
key question was: Would
the active site be wide enough to accomodate the fullerence
derivatives, and allow the two hydrophobic surfaces to contact
these issues was definitely a sideline to the main project I was
working on, but since that project was going slowly, it was easy
to allow my attention to wander over to the fullerenes.
finding myself in front of the graphics workstation, I decided to
finally check if the active site would accommodate a fullerene. The well-known 60-carbon fullerene (C60) has the same
arrangement of pentagons and hexagons as a soccer ball, and it
forms a nearly spherical molecule.
So I began constructing the C60 structure on my computer
using a figure from a paper as a guide.
It was slow
going, trying to keep all the pentagons and hexagons aligned with
each other, but I finally finished and viewed the familiar
spherical shape on the screen.
Then I opened up a protease structure on the same screen
and began moving the fullerene towards the active site.
It was looking
good—the diameter of the fullerene was in the right
ballpark—so I continued nudging the molecules to optimize the
I scrutinized the complex I had created by “hand docking,”
examining thin cross sections for close surface contacts (which
are good) and surface overlaps (which are not).
The outcome was undeniable:
C60 fit almost perfectly into the cylindrical protease
This was quite
clincher, though, came from a “back of the envelope”
calculation estimating of the binding affinity between the
fullerene and the active site. I estimated a binding constant in the nanomolar range, which
would be a very respectable affinity.
The key now was
to find a suitable compound to test. Talk of this project
continued among my friends. Once,
when I was leaving a cafe, I heard some semi-playful jabs about my
I took aside a bright and arrogant friend and asked him,
“Seriously, doesn’t it make sense that fullerenes could be
the right shape, and they're hydrophobic?”
He predicted that
I would never get the fullerenes solubilized. I pointed out that if the fullerene surface had some amino
groups, it should be water soluble.
But he retorted in the voice of the wise old master from Kung Fu, “Ah, Grasshopper, you can't disturb the molecule's
surface without affecting the binding!”
That comment, and his oh-so-certain tone, bothered me even
though I believed that he was wrong.
To jump from
theory to experiment, I needed a fullerene with positively charged
amino groups attached. Not
only would they solubilize the molecule, but they could also bind
to the negatively charged aspartates in the protease active site.
I needed to talk to Fred Wudl at the University of
California at Santa Barbara, who was the only person I knew of who
had pursued adding amino groups to fullerenes.
to Tell the Boss
I had been
working on this side project in my “spare” time and thought it
was time to tell George Kenyon, my thesis advisor.
I went to his office and described the idea, the modeling,
the calculations, and showed him images of the modeled complexes.
He could easily have squelched the project and directed me
to get on with my main project, but instead he said, “Go for
The next day I
described the project in detail to Professor Wudl via e-mail, and
the following day he responded, eager to collaborate. Although he didn’t have the exact compound I'd envisioned,
he did come up with a soluble buckyball with enough of the desired
features that I knew we should try it out.
Diane DeCamp, a
postdoc in the department who was running the HIV assay, ran the
sample. A couple of
weeks later, she came back with the result:
The compound had indeed inhibited HIV protease.
While it wasn’t quite as effective as my
back-of-the-envelope calculation had estimated it would be, it was
still a respectable value for a first compound.
About a year
later, after I had done some more detailed molecular modeling and
enzymology on the compound, the paper on this discovery was
published. It has
attracted quite a bit of interest, being the first biochemical
application of a fullerene. We
are continuing to apply computer-based design techniques to find
tighter binding, more effective compounds.
In the end, I
feel the story illustrates the importance of allowing freedom in
the discovery process: a
freedom to follow a direction others didn’t take seriously, and
a freedom to pursue a project I wasn’t assigned.