et al. show3, amputation of the fin in adult
zebrafish provides a suitable model for studying
angiogenesis during tissue regeneration. Thus,
the combination of (chemo-) genetics and rapid
phenotyping assays promises to provide a pow-
erful new armamentarium for future angiogenic
gene and drug discovery (Fig. 1).
However, critics point out that studying
vascular development in a piscine or amphib-
ian embryo is irrelevant for gaining insight in
angiogenesis in human disease. In addition,
for a new small-animal angiogenesis model
to provide an attractive alternative to the
assays available in mice (the preferred model
of choice for preclinical cancer studies), vessel
growth in this model should resemble angiogen-
esis in mice and humans. Thus, how reliable and
predictive is the adult zebrafish model of regen-
erative angiogenesis developed by Bayliss and
colleagues3? They observed that, in adult zebraf-
ish, blood vessels in the regener-ating fin start to
grow rapidly after fin amputation and initially
form a vascular plexus of fragile channels lined
with naked endothelial cells; these nascent ves-
sels subsequently matured into a more stable
vasculature, in part through coverage of endo-
thelial cells by mural smooth muscle cells3. This
sequence of events is markedly similar to what
occurs in mice and humans1 and thus suggests
a strong resemblance to angiogenesis in mam-
The findings of Bayliss et al. also raise a num-
ber of questions. For instance, tissue regen-
eration in zebrafish relies partly on an initial
dedifferentiation step whereby the injured tis-
sue dedifferentiates to a blastema, which subse-
quently proliferates and gives rise to the different
cell lineages in the regenerating tissue10. Do
endothelial cells in this adult zebrafish model
of fin regeneration also differentiate from
such a primitive blastema, or do they divide
locally from neighboring vessels, as occurs
in mammals? Some previous work suggests
that the latter mechanism may be operating.
Myeloid cells are known to accumulate in the
regenerating fin11—do these hematopoietic cells
also contribute to vessel regeneration? Indeed,
recent studies in mice show that angiocompetent
hematopoietic (stem) cells are mobilized from
the bone marrow, home to sites of active vessel
growth, extravasate and stimulate angiogenesis
by releasing angiogenic factors2. Do endothe-
lial precursors exist in adult zebrafish and are
hematopoietic progenitors mobilized from their
kidney niche (the equivalent of the bone mar-
row niche in mammals)12 to stimulate vessel
regeneration as well? The present regenerating
fin angiogenesis model, in combination with
available protocols for hematopoietic progeni-
tor cell transplantation and the powerful genetic
toolbox available in zebrafish, promises to be
useful for addressing many of these questions.
Bayliss et al. also found that VEGF inhibitors
induce very similar effects in adult zebrafish and
in mice and humans. Indeed, administration of
a VEGF receptor inhibitor inhibited growth
of vessels, especially of the fragile new vessels
with naked endothelial cells, and allowed regen-
eration of only a small avascular region of the
fin3—all reminiscent of what is seen in verte-
brates. Moreover, much as has been observed
in cancer patients treated with VEGF (recep-
tor) inhibitors2, VEGF levels were compensato-
rily upregulated in adult zebrafish treated with
a VEGF receptor inhibitor3. Recent studies in
mice documented that VEGF (receptor) antago-
nists prune pre-existing fenestrated microvessels
by as much as 70%; it will be interesting to see
whether this is also the case in zebrafish. If so,
the zebrafish model will also offer an opportu-
nity to study the safety and toxicity profile of
new anti-angiogenic compounds.
When gene targeting tools in mice became
available some 17 years ago, phenotyping assays
and models had to be developed and miniatur-
ized for use in this small rodent. Skeptics argued
that the mouse would never become a suitable
model for studying certain human disorders,
such as atherosclerosis—a premature critique
that was rapidly silenced when the apoE-defi-
cient mouse was generated. As of 2006, powerful
genetic tools have become available for studying
zebrafish and tadpoles; unfortunately, (adult)
disease models are still largely lacking, permit-
ting the most critical of us to question whether
these primitive animal models will be ever use-
ful for studying angiogenic human disorders.
However, as Bayliss’ pioneering initiative3 shows,
the future of zebrafish and tadpoles for acceler-
ated anti-angiogenic drug discovery may soon
be brighter than was ever expected. Tadpoles
also offer exciting opportunities, as their tail
regenerates after clipping, similar to the fin in
the adult zebrafish, and they have been previ-
ously used for screens of chemical compounds
for cardiovascular development. The prospect of
combining disease models with chemogenetic
approaches in zebrafish and tadpoles should
only further raise our appetite for future fish-
ing and frogging expeditions aimed at locating
1. Carmeliet, P. Nat. Med. 9, 653–660 (2003).
2. Carmeliet, P. Nature 438, 932–936 (2005).
3. Bayliss, P.E. et al. Nat. Chem. Biol 2, 265–273
4. Ny, A. et al. Nat. Med. 11, 998–1004 (2005).
5. Ny, A., Autiero, M. & Carmeliet, P. Exp. Cell Res. 312,
6. Chan, J., Bayliss, P.E., Wood, J.M. & Roberts, T.M.
Cancer Cell 1, 257–267 (2002).
7. Thummel, R. et al. Dev. Dyn. 235, 336–346 (2006).
8. Berghmans, S. et al. Biotechniques 39, 227–237
9. Lawson, N.D. & Weinstein, B.M. Dev. Biol. 248, 307–
10. Poss, K.D. et al. Dev. Biol. 222, 347–358 (2000).
11. Lieschke, G.J., Oates, A.C., Crowhurst, M.O., Ward, A.C.
& Layton, J.E. Blood 98, 3087–3096 (2001).
12. de Jong, J.L. & Zon, L.I. Annu. Rev. Genet. 39, 481–501
Metalloproteases see the light
Small-molecule probes that chemically tag targets by virtue of their enzymatic activities offer a means to focus
system-wide experiments and provide functional information for entire families of proteins. Recent advances in
the design and application of light-activated probes that target metalloproteases have created the opportunity to study
this medically important family of enzymes in unprecedented detail.
Matthew Bogyo is in the Departments of
Pathology and Microbiology and Immunology,
Stanford University, 300 Pasteur Drive,
Stanford, California 94305, USA.
“In the middle of every difficulty lies opportu-
nity.” This statement, made by Albert Einstein
decades ago, seems to perfectly address the
issues at the heart of the field of proteomics.
Researchers in this area continue to face great
challenges and difficulties in finding ways to
monitor potentially hundreds of thousands of
proteins that are present at levels that can span
as many as six orders of magnitude in any given
proteome1. Yet a system-wide understanding
NEWS AND VIEWS
NATURE CHEMICAL BIOLOGY VOLUME 2 NUMBER 5 MAY 2006
© 2006 Nature Publishing Group http://www.nature.com/naturechemicalbiology
of protein regulation and function provides
vast opportunities, ranging from identification
of new therapeutic targets and biomarkers for
human disease to mapping of the fundamen-
tal pathways of cellular survival. New tools and
methodologies that enable meaningful pro-
teomic studies are the keys to helping this field
realize these opportunities. Motivated by this,
Cravatt and co-workers describe their global
look at a key enzyme superfamily in this issue
of Nature Chemical Biology2.
Arguably one of the greatest advances in
our ability to globally monitor protein regula-
tion and function has been the development
of methods to enrich samples for subsets of
proteins or protein families, thereby reducing
problems associated with sample complexity.
One widely used method adopts small-
molecule tags that chemically modify protein
targets, allowing them to be rapidly isolated
for proteomic analysis. The ability to selectively
enrich samples for proteins of interest is particu-
larly relevant because important regulatory pro-
teins such as transcription factors and enzymes
are often expressed at exceedingly low levels1.
Thus studies of families of enzymatic proteins
such as proteases require the development of
probes that can selectively fish out these targets
from the sea of other proteins.
A second important issue facing proteomic
research is the need to devise ways to moni-
tor not just the expression levels but also the
functional regulation of proteins. Virtually
all global proteomic methods only provide
information about overall abundance, yet
most proteins involved in critical biological
processes are regulated through tightly con-
trolled post-translational mechanisms. For
example, proteases are synthesized as inactive
zymogens that must be activated in a spatially
and temporally controlled manner. Therefore,
the ability to monitor the dynamics of the
regulation of their activity is necessary for
understanding their function in both normal
and disease processes.
Cravatt and co-workers describe the synthe-
sis and application of a small library of light-
activated probes that specifically target the active
forms of metalloproteases2. These reagents
represent an important addition to an ever-
growing toolbox of reagents known as activity-
based probes (ABPs) that both enable enrich-
ment of a specific target class of proteins and
provide dynamic information on the regulation
of their enzymatic activity. Although a number
of ABPs that target proteases have found wide-
spread applications in studies of protease func-
tion (for reviews see refs. 3–6), this technology
has remained difficult to apply to metallopro-
teases owing to the fact that these enzymes do
not form stable covalent bonds with a substrate
(that is, acyl enzyme intermediates) during
catalysis. Most ABPs make use of chemically
reactive functional groups that are capable
of specific modification of reactive nucleo-
philes within the enzyme active site (Fig. 1).
In the case of serine and cysteine proteases, this
nucleophile is a catalytic hydroxyl or thiol and
the reaction results in stable chemical linkage
to the enzyme. For metalloproteases the pri-
mary catalytic nucleophile is a bound water
molecule. Therefore ABPs that target this fam-
ily of enzymes must form stable linkages with
potentially unreactive amino acid residues in or
around the active site of the enzyme.
Cravatt and co-workers have solved this prob-
lem by developing metalloprotease ABPs carry-
ing a light-activated cross-linker that facilitate
permanent covalent modification of target
proteases upon exposure to UV light (Fig. 1).
Because the probes bind only when a protease
active site is properly formed and free of inhibi-
tors, the resulting labeling provides an indirect
indication of the levels of active proteases within
a sample7. By making small libraries of primary
peptide sequences, the authors show that it is
possible to identify probe sets that effectively
target a wide range of metalloprotease targets.
So why target metalloproteases? Proteases
make up nearly 2% of the human genome8
and represent 5–10% of all known potential
drug targets9. In particular, there are nearly 200
distinct metalloproteases whose functions are
linked to a number of clinically relevant condi-
tions, most notably cancer. For this reason many
large pharmaceutical companies initiated clini-
cal trials with matrix metalloprotease inhibi-
tors for treatment of late-stage cancer patients.
Unfortunately, none of these trials produced
promising results and most have now been
abandoned9. It is becoming clear that the tri-
als may have failed in part because of a lack of
understanding of the complex functional roles
of metalloproteases in disease progression. Thus
ABPs such as the ones described in this issue are
likely to provide valuable new information that
could aid future attempts to target this family of
proteases for therapeutic gain.
1. Ghaemmaghami, S. et al. Nature 425, 737–741
2. Sieber, S. A., Niessen, S., Hoover, H. S. & Cravatt, B. F.
Nat. Chem. Biol. 2, 274–281 (2006).
3. Berger, A.B., Vitorino, P.M. & Bogyo, M. Am. J.
Pharmacogenomics 4, 71–81 (2004).
4. Jessani, N. & Cravatt, B.F. Curr. Opin. Chem. Biol. 8,
5. Speers, A.E. & Cravatt, B.F. ChemBioChem 5, 41–47
6. Jeffery, D.A. & Bogyo, M. Curr. Opin. Biotechnol. 14,
7. Saghatelian, A., Jessani, N., Joseph, A., Humphrey, M.
& Cravatt, B.F. Proc. Natl. Acad. Sci. USA 101, 10000–
8. López-Otín, C. & Overall, C.M. Nat. Rev. Mol. Cell Biol.
3, 509–519 (2002).
9. Overall, C.M. & Kleifeld, O. Nat. Rev. Cancer 6, 227–
Figure 1 Mechanisms of protease active site labeling by activity-based probes. (a) Probes that target
cysteine proteases make use of the catalytic thiol residue to form a permanent covalent bond (dashed box)
between the probe and target enzyme. (b) Metalloproteases do not use a catalytic amino acid side chain as
the primary nucleophile. Therefore, probes bind the catalytic zinc ion but require a light-activated cross-
linker to chemically modify (dashed box) the target enzyme.
NEWS AND VIEWS
VOLUME 2 NUMBER 5 MAY 2006 NATURE CHEMICAL BIOLOGY
© 2006 Nature Publishing Group http://www.nature.com/naturechemicalbiology