news and views
lymphatic vessels in the corneas of sVEGFR-2–
deficient mice at birth1. In the adult mice, lack
of sVEGFR-2 resulted in more lymphangio-
genesis in corneas after injury. Surprisingly,
these corneas were not invaded by blood ves-
sels. One reason for this specificity is that only
dimeric VEGFR-2 can bind VEGF-A and block
angiogenesis5, whereas sVEGFR-2 is a mono-
mer in the cornea. The authors proposed that
sVEGFR-2 prevents lymphatic growth in the
cornea because it is a selective VEGF-C antago-
The absence of lymphatic vessels in the cornea
grants the tissue a uniquely immune-privileged
status; it prevents traffic of antigen-presenting
cells into the lymph nodes and induction of
immune response, and so it prevents graft
rejection6. Inflammation, however, can induce
lymphangiogenesis and lead to the loss of this
immune privilege, rendering corneal allografts
much more susceptible to rejection.
Because blocking lymphangiogenesis can pre-
vent graft rejection, Albuquerque et al.1 tested
the effects of sVEGFR-2 in corneal transplan-
tation in mice. They found that administering
sVEGFR-2 into the cornea inhibited lymphatic
vessel formation and doubled transplant sur-
vival rate, and they ascribed this effect to the
blockade of lymphangiogenesis1.
The authors acknowledge that sVEGFR-2
could promote allograft survival also by inhibit-
ing dendritic cell trafficking to the lymph nodes7.
Nevertheless, their study is in line with others
that found lymphangiogenesis and VEGF-C to
be important in transplant immunobiology7–9.
Notably, chronic rejection of human kidney
transplants is also associated with lymphangio-
genesis9. The above studies provide a basis for
evaluating the effects of sVEGFR-2 and other
inhibitors of lymphangiogenesis on prolonging
the life of various organ transplants.
Outside of the eye, the authors found
sVEGFR-2 expression to be particularly abun-
dant in the epidermis1. Removing sVEGFR-2
in the mouse epidermis caused lymphatic ves-
sel hyperplasia, further arguing that sVEGFR-2
antagonizes VEGF-C. Albuquerque et al.1 also
report that endothelial cells from blood vessels
secrete sVEGFR-2. sVEGFR-2 might, specu-
latively, direct the proper separation of blood
vessels from lymphatic vessels.
The identification of sVEGFR-2 as an endog-
enous, specific inhibitor of lymphangiogenesis
is a major contribution to the field and comple-
ments research efforts by other labs to decouple
blood and lymphatic vessel growth. This work,
however, may also spark controversy, and addi-
tional studies will be required to confirm the
selectivity of sVEGFR-2 for lymphangiogenesis.
It is surprising that the authors did not report
any effects of sVEGFR-2 on angiogenesis,
because VEGF-C is a potent chemotactic factor
for macrophages10, which are a rich source of
angiogenic factors. Macrophages infiltrate the
cornea during inflammation6,11, attracted by
VEGF-C itself, and so blocking of VEGF-C with
sVEGF-2 would be expected to suppress corneal
angiogenesis at least to a certain degree.
The work by Albuquerque et al.1 could make
a substantial contribution to the burgeoning
field of lymphangiogenesis in various diseases.
In cancer, the growth of lymphatic vessels can
promote metastasis2,12, but whether endog-
enous inhibitors of lymphangiogenesis could
affect tumor progression is unknown. It will be
exciting to see whether sVEGFR-2 is expressed
in tumors. The finding that sVEGFR-2 inhib-
its lymphangioma proliferation in vitro1 raises
the possibility of its use in treatment of lymph-
angioma and other lymphatic vascular mal-
formations. The discovery of sVEGFR-2 may
even offer a fresh perspective for looking at the
underlying cause of lymphedema, in which an
endogenous inhibitor could hamper lymphatic
regeneration. Finally, we are just beginning to
learn about the interface of the lymphangio-
genesis and adaptive immune responses. These
research efforts have potential implications for
modulation of antitumor immunity in addition
to the allorejection immunity described above.
But what about the eye? This new tool could
be used therapeutically to improve survival of
corneal transplants, and it may prove useful for
treatment of many immune-mediated destruc-
tive corneal diseases characterized by aberrant
lymphangiogenesis. Let’s see!
1. Albuquerque, R.J.C. et al. Nat. Med. 15, 1023–1030
2. Alitalo, K., Tammela, T. & Petrova, T.V. Nature 438,
3. Ambati, B.K. et al. Nature 443, 993–997 (2006).
4. Cursiefen, C. et al. Proc. Natl. Acad. Sci. USA 103,
5. Fuh, G., Li, B., Crowley, C., Cunningham, B. & Wells,
J.A. J. Biol. Chem. 273, 11197–11204 (1998).
6. Patel, S.P. & Dana, R. Semin. Ophthalmol. 24, 135–138
7. Chen, L. et al. Nat. Med. 10, 813–815 (2004).
8. Cursiefen, C. et al. Invest. Ophthalmol. Vis. Sci. 45,
9. Kerjaschki, D. et al. Nat. Med. 12, 230–234 (2006).
10. Skobe, M. et al. Am. J. Pathol. 159, 893–903 (2001).
11. Maruyama, K. et al. J. Clin. Invest. 115, 2363–2372
12. Skobe, M. et al. Nat. Med. 7, 192–198 (2001).
volume 15 | number 9 | september 2009 nature medicine
Two sides to cilia in cancer
The primary cilium can keep cancer at bay, or it can instigate tumor development, according to studies in mice
(pages 1055–1061 and 1062–1065). The outcome depends on the nature of the initiating event, which involves
signaling through the Hedgehog pathway.
Rune Toftgård is at the Karolinska Institutet Center
for Biosciences, Department of Biosciences and
Nutrition, NOVUM, Huddinge, Sweden.
Primary cilia, present on most cells in the body
during interphase, were first reported more
than 100 years ago1. This cellular organelle
consists of a microtubule core connected to a
basal body derived from the centriole, a com-
ponent of the mitotic spindle; in most cases,
the cilium is immotile. Cilia serve a number
of key functions in cell-to-cell and cell-to-
environment communication, ranging from
mechanosensation to the transduction of key
developmental signaling pathways. Studies of
a group of human hereditary disorders, collec-
tively termed ciliopathies, have linked muta-
tions in cilia-associated proteins to a variety of
abnormalities, such as reversed asymmetry of
body organs (situs inversus), polydactyly, cystic
kidney and others.
In this issue of Nature Medicine, two studies
report that cilia seem to have opposite effects
on tumorigenesis depending on the context:
cilia are either required for or can repress tum-
origenesis2,3. Cilia are central to the activity of
the Hedgehog pathway, and the point at which
oncogenic mutations disrupt the cascade may
determine how the tumor responds to cilia dis-
ruption (Fig. 1).
The Hedgehog pathway, a key player in nor-
mal development, is dysregulated in a number
© 2009 Nature America, Inc. All rights reserved.
news and views
nature medicine volume 15 | number 9 | september 2009
have lower expression levels of the protein,
expression of GLI2 was not sufficient to induce
medulloblastomas2. Moreover, disruption of
cilia by itself did not influence the amount of
the repressor Gli3 in skin keratinocytes3. These
experiments suggest the cilia may regulate
more targets and signaling pathways to pro-
mote tumor formation. Alternatively, perhaps
the absence of cilia changes the state of differ-
entiation of target cell populations or loosens
the restraints on cell cycle progression. Future
experiments, such as the conditional deletion
of Gli3, may show whether the Hedgehog path-
way is the primary mechanism by which cilia
repress tumor formation.
The finding that cilia may repress tumor for-
mation is not totally unexpected, given previous
studies showing that the von Hippel–Lindau
(VHL) tumor suppressor maintains primary
cilia. When renal cells have VHL mutations,
they form renal cysts that lack cilia, the pre-
cursors of clear cell renal cell carcinoma. The
increased epithelial cell proliferation in the
cysts suggests that primary cilia may impose
cell cycle restrictions6,7.
The mechanistic insight gained from these
mouse models of BCC and medulloblas-
toma is probably relevant to human disease.
Confirming an earlier report8, Wong et al.3
detected ciliated cells in most BCCs. This
observation is consistent with observations
that essentially all BCCs harbor gene muta-
tions that activate the Hedgehog pathway at
the receptor level. The situation is murkier in
models. Both groups induced tumorigenesis
by expressing an activated SMO protein in
mice that lacked a kinesin required for cilium
formation. They found that, as expected, dis-
ruption of primary cilia led to a strong inhi-
bition of both tumor types2,3. When the cilia
were disrupted, SMO could no longer induce
the expression of Hedgehog target genes or
the cell proliferation, which led to tissue
hyperplasia. These results confirmed that cilia
are necessary for oncogenic Hedgehog sig-
naling as well as for signaling during normal
The researchers then attempted to under-
stand at which step the Hedgehog pathway
was dependent on cilia, expressing an active
GLI2 transcription factor while disrupting the
structure of the cilia2,3. In this situation, Han
et al.2 and Wong et al.3 observed a marked
acceleration of tumor formation accompanied
by increased expression of Hedgehog target
genes implying that, unexpectedly, absence of
cilia may enhance activation of the Hedgehog
signaling pathway. The mechanism underlying
the increased Hedgehog activity in tumor cells
lacking cilia may lie in the balance between the
activity of transcriptional activator GLI2 and
that of the repressor form of GLI3. Low levels of
repressor GLI3, perhaps through impaired pro-
cessing, may leave the active GLI2 transcription
factor unopposed to promote cell proliferation
and stem cell renewal for tumor growth.
Some questions remain. In mice heterozy-
gous for a gene encoding a mutant Gli3, which
of cancers, including basal cell carcinoma
(BCC) of the skin, medulloblastoma and pan-
creatic cancer. BCC, the most common cancer
in individuals of European ancestry, is benign
in nature and invariably harbors an activated
Hedgehog signaling pathway. Similarly, about
25%–30% of medulloblastomas show active
Hedgehog signaling. This primitive neuroec-
todermal tumor of the cerebellum is the most
common malignant brain tumor in children,
with current treatment options resulting in
a five-year survival rate on the order of 60%
In mammals, there are three secreted ligands
in the Hedgehog family: Sonic hedgehog,
Indian hedgehog and Desert hedgehog. All
of these interact with the Patched receptor
protein (PTCH). In the absence of Hedgehog
binding, PTCH represses signal transduction
by inhibiting the seven-transmembrane pro-
tein co-receptor Smoothened (SMO). Upon
binding of ligand to PTCH, the inhibition of
SMO is relieved, resulting in the translocation
of the transcription factors GLI1 and GLI2 to
the nucleus. In the nucleus, GLI1 and GLI2 act
primarily as activators to turn on target genes,
providing feedback control of the pathway and
regulating cell proliferation and self-renewal.
(In the absence of Hedgehog signaling, GLI3,
processed into a repressor form, predominates
in the nucleus; it is a weak activator as a full-
Cilia are crucial for the proper function of
the Hedgehog signaling pathway, although the
detailed mechanisms still remain unclear. It is
known that intraflagellar transport (IFT) pro-
teins, which transport proteins to and from the
tip of primary cilia, are required for Hedgehog
signaling activity5. Moreover, Hedgehog sig-
naling pathway components dynamically traf-
fic throughout the primary cilium.
When Hedgehog ligand is present, PTCH
moves out of the cilium and, together with the
ligand, is internalized through endosomes into
the cytoplasm. At the same time, the SMO co-
receptor is activated and travels from intra-
cellular stores to the cilium, most likely via an
IFT-dependent process, allowing the produc-
tion of activator GLI2 and preventing the pro-
duction of repressor GLI3. Such observations
suggest that ligand- and receptor-dependent
Hedgehog signaling requires the presence of
primary cilia. A caveat to this model is that a
role of IFT proteins in regulation of Hedgehog
signaling at cellular locations other than pri-
mary cilia cannot yet be excluded.
Basal cell carcinomas and medulloblast-
omas are heavily dependent on Hedgehog sig-
naling. Given such observations, Han et al.2
and Wong et al.3 investigated how cilia influ-
ence the formation of these tumors in mouse
activated at receptor level
Hedgehog pathway Hedgehog pathway
of receptor level
Limited cell proliferation
Suppressed tumor formation
Accelerated tumor formation
Figure 1 Dual role of primary cilia in Hedgehog (Hh)-induced tumor development. (a) In the absence
of ligand, the PTCH receptor is localized to the cilia, whereas the SMO co-receptor is excluded from
the cilia. The GLI3 repressor (GLI3R) prevents activation of target genes. (b) In the presence of primary
cilia, the mutant oncogenic Hedgehog co-receptor SMO constitutively localizes to this organelle,
inhibits formation of GLI3R and activates the downstream GLI transcription factors (for example, GLI2).
GLI2 stimulates target gene activation, promoting increased cell proliferation and tumor development.
(c) Activation of GLI2 can also preferentially stimulate tumor development in the absence of cilia. The
authors propose that GLI3R is downregulated in these cells, but the mechanism is still unclear2,3.
© 2009 Nature America, Inc. All rights reserved.
news and views
volume 15 | number 9 | september 2009 nature medicine
tests or therapies targeting ciliary function for
treatment of BCC or medulloblastoma.
1. Zimmerman, K.W. Arch. Mikrosk. Anat. 52, 552–706
2. Han, Y.-G. et al. Nat. Med. 15, 1062–1065 (2009).
3. Wong, S.Y. et al. Nat. Med. 15, 1055–1061 (2009).
4. Polkinghorn, W.R. & Tarbell, N.J. Nat. Clin. Pract.
Oncol. 4, 295–304 (2007).
5. Huangfu, D. et al. Nature 426, 83–87 (2003).
6. Thoma, C.R., Frew, I.J. & Krek, W. Cell Cycle 6, 1809–
7. Plotnikova, O.V., Golemis, E.A. & Pugacheva, E.N.
Cancer Res. 68, 2058–2061 (2008).
8. Wilson, R.B. & McWhorter, C.A. Lab. Invest. 12, 242–
9. Northcott, P.A. et al. Nat. Genet. 41, 465–472
10. Taylor, M.D. et al. Nat. Genet. 31, 306–310 (2002).
11. Jia, J. et al. Dev. Biol. 330, 452–460 (2009).
12. Seeley, E.S., Carrière, C., Goetze, T., Longnecker, D.S.
& Korc, M. Cancer Res. 69, 422–430 (2009).
13. Olive, K.P. et al. Science 324, 1457–1461
Perhaps hints about the potential for thera-
peutics can be found in pancreatic cancer,
where Hedgehog signaling is also frequently
activated. Most pancreatic tumor cells are
unciliated12, potentially as a result of preva-
lent KRAS oncoprotein mutations. Tumor
cells express and secrete Hedgehog ligand, but
an autocrine response is absent. Instead the
cells of the tumor stroma, which frequently
have cilia, respond to the Hedgehog signal in
a paracrine manner. Recently, a study found
that SMO antagonists could deplete tumor
stroma13, implying the importance of cilia for
the Hedgehog pathway for these cells.
Despite these advances, deeper understand-
ing of how primary cilia operate in Hedgehog
signaling and other pathways is urgently
needed before the introduction of diagnostic
medulloblastoma, however, because activation
of the Hedgehog signaling pathway down-
stream of the receptor may occur in some
tumors9,10. In tumor tissue samples, Han et
al.2 found that ciliated tumor cells had either
active Hedgehog or Wnt signaling pathways,
which is a suggestive correlation but one that
needs confirmation in larger studies.
Activation of the Hedgehog pathway down-
stream of the receptor may select against the
presence of cilia in a tumor, and recent observa-
tions show that the relative level of Gli3 repres-
sor can be lowered independently of cilia11. In
light of these observations, the diagnostic value
of cilia as a biomarker for targeted therapy
with SMO antagonists will require assessment
of Hedgehog pathway activity in addition to
Connecting obesity, aging and diabetes
Rexford S Ahima
Obesity accelerates the aging of adipose tissue, a process only now beginning to come to light at the molecular level.
Experiments in mice suggest that obesity increases the formation of reactive oxygen species in fat cells, shortens
telomeres—and ultimately results in activation of the p53 tumor suppressor, inflammation and the promotion of
Rexford S. Ahima is in the Department of
Medicine, Division of Endocrinology, Diabetes and
Metabolism, the Institute for Diabetes, Obesity and
Metabolism, University of Pennsylvania School of
Medicine, Philadelphia, Pennsylvania, USA.
As technology has improved hygiene, the food
supply and living standards overall, there has
been a rise in such age-related illnesses as
cardiovascular disease, cancer, degenerative
diseases of the brain and other organs, and met-
abolic disorders such as diabetes. Age-related
disorders have become widespread throughout
the world, replacing infectious diseases as the
leading cause of death in developed countries.
As we age, many people develop the metabolic
syndrome, characterized by central (visceral)
obesity, insulin resistance, impaired glucose
tolerance or overt diabetes, hypertension, dys-
lipidemia and cardiovascular complications.
Diabetes is also a recognized cause of accel-
erated aging, but the mechanisms linking dia-
betes and aging are not well understood. Work
from Minamino et al.1 in this issue of Nature
Medicine offers insights into how obesity affects
the aging of adipose tissue, influencing inflam-
mation and glucose homeostasis.
Obesity is a major cause of insulin resis-
tance, which progresses to type 2 diabetes when
the pancreas is unable to produce sufficient
amounts of insulin. In recent years, evidence has
emerged that inflammation has a crucial role in
the development of insulin resistance, diabetes
and cardiovascular diseases associated with obe-
sity2,3. Macrophages infiltrate adipose tissue in
obese states, and cytokines levels are elevated
and cause insulin resistance and diabetes2,3.
The deterioration of the structure and
function of organs during aging is associ-
ated with oxidative stress, genetic instability
and disruption of homeostatic pathways4.
Much aging research has studied telomeres,
which are composed of tandem repeats of the
TTAGGG sequence and associated proteins
and are located at the ends of chromosomes5.
Stem cells and cancer cells are able to continue
dividing because the telomeres are maintained
by an enzyme called telomerase. In contrast, in
normal somatic cells, the telomeric repeats are
lost with each cell cycle until a ‘critical length’ is
attained. The shortening of telomeres leads to
activation of tumor suppressors, in particular
p53, which induces cell cycle arrest and aging.
Genomic damage can also accrue over time
from reactive oxygen species (ROS). Diabetes,
among other age-related illnesses, is associated
with an inability to detoxify ROS6. Similarly,
telomere shortening has been linked to obe-
sity, insulin resistance, diabetes and coronary
Because there are similarities in metabolic
dysregulation in aging and obesity, it is likely
these conditions share similar cellular path-
ways. To test this hypothesis, Minamino et al.1
analyzed the adipose tissue of obese mice for
evidence of oxidative stress, aging and inflam-
mation. Adipose tissue from agouti mice,
which are genetically obese, had higher lev-
els of ROS and DNA damage than lean mice
when both groups were on a normal diet for
20 weeks1. Adipose tissue from agouti mice
showed features of premature aging, such as
a higher expression of senescence-associated
β-galactosidase, p53 and cyclin-dependent
kinase inhibitor 1A (Cdkn1a).
The researchers then determined whether
age-related changes in adipose tissue were
responsible for insulin resistance1. Adipose
tissue from agouti mice also expressed proin-
flammatory cytokines that attract macrophages
to adipose tissue, tumor necrosis factor and
monocyte chemoattractant protein-1, which are
associated with insulin resistance. Adiponectin,
which enhances insulin action, was suppressed
in adipose tissue from agouti mice.
© 2009 Nature America, Inc. All rights reserved.