NEWS AND VIEWS
the presence or absence of alleles is reported.
2004 Nature Publishing Group http://www.nature.com/naturegenetics
showed no significant deviation from Hardy- HapMap project12, needs to cover segmental
For a duplicon variant, this leads to confound-
ing of one of the homozygous classes with the
heterozygous class (Table 1). MSVs will be
scored as heterozygous if there is variation in
the sample at any interrogated site. Such geno-
type misassignment has the potential to create
uncertainties in mapping studies.
The quantification required to genotype
these polymorphisms correctly is tedious, and
not possible with all genotyping methodolo-
gies13. In any case, multisite polymorphisms
are difficult to interpret. Thus, other
approaches to identify duplicon markers in
an initial screen are attractive. Segregation
may be inconsistent with expectations of an
ordinary SNP. If insufficient family samples
are available to detect this, an alternative is to
examine deviations from Hardy-Weinberg
equilibrium. But Fredman et al. show
instances with little or no deviation from
Hardy-Weinberg when allele presence or
absence is scored. Only 50% of the MSVs that
they identified showed significant deviation
(P < 0.01) when tested in 32 diploid DNA
samples. To counter the argument that 32
may be too small for sufficient power, they
provide the example of two MSVs that
TGF signaling in health and disease
Rosemary J Akhurst
Inactivating mutations in TGFBR2, encoding the transforming growth factor- (TGF) type 2 receptor, may account
for up to 10% of cases of Marfan syndrome. This finding has implications for a wider spectrum of disorders,
including cancer, fibrosis and inflammatory and cardiovascular diseases, which are influenced by TGF.
Marfan syndrome is a dominantly inherited
disorder of fibrous connective tissue affect-
ing the skeletal, cardiovascular and ocular
systems, with an incidence of 1 in 10,000.
Some have suggested that Abraham Lincoln
and Paganini had Marfan syndrome.
Approximately 80% of Marfan cases are due
to inactivating mutations in FBN1, encod-
ing the extracellular microfibrillar protein
fibrillin. Controversy surrounds the issue of
clinical and genetic heterogeneity in the
remaining 20% of individuals with Marfan
syndrome, specifically with respect to a sec-
ond locus on chromosome 3p24.2p25
(refs. 1,2). On page 855 of this issue,
Rosemary J. Akhurst is at the Mt. Zion Cancer
Research Institute, University of California, San
Francisco, USA. e-mail: rakhurst@cc.ucsf.edu
790
Weinberg equilibrium in more than 1,500
individuals. This leaves their recommended
procedure: incorporating CHM samples sys-
tematically as controls in the genotyping
experiments.
Although MSVs may be avoided as markers
when possible, there are instances where it is
desirable to interrogate a single site in the
presence of complex polymorphism.
Examples might include the TCR/ and
IGH loci on chromosome 14, which have
complicated repeat structures and potential
disease relevance. Other examples occur in
plants: Brassica napus (rapeseed) is an
amphidiploid, composed of copies of B. oler-
acea (cauliflower) and B. rapa (turnip)
genomes. Four copies of the same gene may
be present, two originating from each of the
ancestral species. To be sure of obtaining spe-
cific SNP genotyping, a PCR step can be
designed to amplify a single gene followed by
allele-specific detection. Fredman et al. spec-
ulate that MSVs might in some instances have
direct relevance to disease and, therefore, be
important to characterize.
A complete linkage disequilibrium map of
the human genome, such as proposed in the
Takeshi Mizuguchi and colleagues3 report
that inactivating mutations in TGFBR2 at
3p24.2p25 cause Marfan syndrome type 2
(MFS2). This finding may fuel further
debate as to the molecular etiology of this
disease, as it has been suggested that fib-
rillin functions by limiting activation of
TGF in vivo4.
TGFs and cancer
The TGFs are multifunctional growth fac-
tors that signal through a well-characterized
pathway of transmembrane serine-threo-
nine kinase receptors and intracellular sig-
naling molecules of the SMAD family5.
Their biological effects are context-depen-
dent, varying with tissue type and the activ-
ity of other intracellular signaling pathways
in the cell. They are potent negative growth
VOLUME 36 | NUMBER 8 | AUGUST 2004 NATURE GENETICS
duplications. Although this may require spe-
cial attention to duplicon variation and geno-
typing, Fredman et al. reach the conclusion
that 50% of the markers in the public data-
base for these regions are probably true SNPs,
a density similar to that found elsewhere in
the genome. The concentration of genes in
duplicated segments that they and others3
report shows that it will be important to have
these SNPs deployed for disease-mapping.
1. Fredman, D. et al. Nat. Genet. 36, 861866 (2004).
2. Bailey, J.A. et al. Science 297, 10031007 (2002).
3. Cheung, J. et al. Genome Biol. 4, R25 (2003).
4. Emanuel, B.S. & Shaikh, T.H. Nat. Rev. Genet. 2,
791800 (2001).
5. Shaw, C.J. & Lupski, J.R. Hum. Mol. Genet. 13 Spec
No 1, R57R64 (2004).
6. Samonte, R.V. & Eichler, E.E. Nat. Rev. Genet. 3,
6572 (2002).
7. Estivill, X. et al. Hum. Mol. Genet. 11, 19871995
(2002).
8. Hurles, M. Science 298, 1509 (2002).
9. Prince, J.A. et al. Genome Res. 11, 152162 (2001).
10. Schouten, J.P. et al. Nucleic Acids Res. 30, e57
(2002).
11. Sherry, S.T. et al. Nucleic Acids Res. 29, 308311
(2001).
12. The International HapMap Consortium. Nature 426,
789796 (2003).
13. Gut, I.G. in Molecular Analysis and Genome Discovery
(eds. Rapley, R. & Harborn, S.) 4369 (Wiley,
Chichester, UK, 2004).
regulators and can also induce differentia-
tion, apoptosis, cell migration, adhesion
and extracellular matrix deposition5. In a
Franco-Japanese collaboration, Mizuguchi
et al.3 investigated 11 sporadic and 10 famil-
ial cases of Marfan syndrome not associated
with mutations in FBN1. They found a com-
plex cytogenetic disruption of TGFBR2 in
one individual with Marfan syndrome, as
well as four independent missense muta-
tions, two of which segregated with the dis-
ease in family studies, and one of which
occurred in both a French family and an
independent Japanese individual. All the
missense mutations occur at conservative
amino-acid residues in the serine-threonine
kinase domain of the receptor, inactivating
the receptor, as assayed by TGF reporter
assays in vitro.
 NEWS AND VIEWS

It may come as a surprise to cancer biolo- Table 1 Germline mutations in components of the TGF superfamily signaling pathways
gists that individuals with MFS2 do not
have a higher rate of cancer. TGFBR2 has Gene OMIM OMIM
been reported to be a tumor-suppressor number Syndrome number

gene, particularly for cancers of the gas- ENG 131195 HHT1 187300
trointestinal tract5. Downregulation of ACVRL1 601284 HHT2 600376
TGF1 or TRII (the TGF type II recep- SMAD4 600993 JPS 174900
tor) in transgenic mice has been shown to JPS and HHT syndrome 175050
enhance epithelial tumor development5, BMPR1A 601299 JPS 174900
and genetic ablation of Tgfbr2 in mouse BMPR2 60079 Familial primary pulmonary hypertension 1 178600
2004 Nature Publishing Group http://www.nature.com/naturegenetics

fibroblasts results in development of TGFB1 190180 Camurati-Engelmann disorder 131300

aggressive carcinoma of the stomach and
prostate6. It is possible that individuals with
MFS2 die from other complications before
cancer can take hold, namely aortic rupture
due to weakening and dilatation of the
ascending aorta, a characteristic of Marfan
syndrome. But median survival of individ-
uals with Marfan syndrome is 53 years for
men and 72 years for women, and one of
the four families with MFS2 that were stud-
ied included affected members who lived
well past 60 years (C. Boileau, personal
communication). Moreover, germline
mutations in other components of the
TGF pathway, specifically SMAD4 and
BMPR1A, lead to childhood development
of multiple intestinal polyps and early-
onset stomach cancer in the form of juve-
nile polyposis syndrome (JPS; Table 1).
Although a germline TGFBR2 mutation has
previously been reported to result in hered-
itary nonpolyposis colorectal cancer,
Mizuguchi et al.3 found that this same
amino-acid change occurs in 7% of the
normal population and is not associated
with incidence of sporadic colorectal can-
cer, suggesting that it is a rare functional
polymorphism in TGFBR2 rather than a
cancer-causing mutation.
Perhaps individuals with Marfan syndrome
are even protected from developing life-
threatening invasive carcinoma. Although
TGF is a potent growth inhibitor, elevated
TGF signaling in tumors contributes to car-
cinoma progression and metastasis, and indi-
viduals with colon carcinomas lacking
TGFBR2 genes have a better prognosis than
those that maintain this signaling pathway
intact5. It will certainly be reassuring to phar-
maceutical companies developing smart
drugs designed to inhibit TGF receptor
kinases that haploinsufficiency with respect
to TGFBR2, even from the earliest stages of
development, does not result in increased
malignancy later in life.
Simple to complex
Common themes of germline mutations in
components of the TGF signaling pathway
(Table 1) are that they invariably involve
TGFBR2 190182 MFS2 154705
FBN1 134797 MFS1 154700
FBN1 is included although it is not a signaling component. It is an extracellular matrix protein that may influence
TGF signaling by acting as a pool or activating matrix for the ligand4.
inactivating mutations resulting in autosomal TGF signaling pathway is their variable
dominant disorders and that they mainly expressivity. The phenotypic spectrum of
affect the skeletal and cardiovascular systems. each of these single-gene defects is heteroge-
There is also considerable overlap of pheno- neous, even between members of the same
types. Camurati-Engelmann syndrome, a family, which can lead to confusion in clini-
bone overgrowth defect due to TGFB1 hap- cal classification1,2. Disease severity and the
loinsufficiency, has several Marfan syn- relative involvement of different organ sys-
dromelike features, including long and tems are influenced by environmental fac-
slender extremities. Mutations of either ENG tors and modifier genes. One of these
or ACVRL1, encoding a type III and a type I genetic modifiers may be TGFB1 itself. A
receptor, respectively, cause hereditary hem- missense polymorphism in the signal pep-
orrhagic telangiectasia (HHT), a late-onset tide, L10P, has been associated with
vascular dysplasia with breakdown of capillary increased systemic TGF1 levels. Forty per-
structure and tortuous arterio-venous shunts cent of the population is heterozygous with
that can result in vascular rupture. Mutations respect to this polymorphism, and posses-
in SMAD4 may also lead to HHT together sion of the proline variant is genetically
with JPS. Finally, primary pulmonary hyper- associated with altered risk for a number of
tension, which is due to precapillary blockade cardiovascular and fibrotic diseases (Table
of pulmonary arteries, can be caused by muta- 2). Genetic modifier loci that confer devel-
tions in BMPR2 and has also been observed in opmental redundancy for Tgfb1 in the
individuals with HHT7. It remains to be seen mouse may also influence phenotypic out-
whether familial thoracic aortic aneurysms come in this group of diseases9. Human and
and dissections 2, a syndrome distinct from mouse genetic studies have already taught
MFS2 but with similar cardiovascular mani- us much about the biological function of the
festations and mapping to 3p2425 (ref. 8), is TGF signaling pathway in vivo. Further
caused by mutations in TGFBR2. genetic studies will clarify this network of
Another feature of syndromes caused by molecular interactions that is important in
germline mutations in components of the so many human diseases.
Table 2 Polymorphisms in TGFB1 modify risk for cardiovascular and fibrotic diseases
Variant amino acid Population size
conferring high risk/ Circulating (total or affected/
Disease increased severity TGF1 level unaffected) Reference
Myocardial infarction Pro25 Low 561/629 10
Hypertension/blood pressure Arg25 High 1,190 10
End-stage cardiomyopathy Pro10 (not Arg25) High 253/94 11
Proliferative diabetic retinopathy Arg25 High 73/172 NIDDMs 12
Arthritis Leu10 Low 155/110 13
Systemic sclerosis Pro10 High 149/147 14
Hepatic fibrosis Pro10 High 48/97 15
NIDDMs, noninsulin-dependent diabetes mellitus.

NATURE GENETICS VOLUME 36 | NUMBER 8 | AUGUST 2004 791

 NEWS AND VIEWS

1. Dietz, H. et al. Nat. Genet. 9, 228229 (1995). (2004). 979984 (2001).

2. Gilchrist, D.M. Am. J. Hum. Genet. 54, 553554
(1994).
3. Mizuguchi, T. et al. Nat. Genet. 36, 855860 (2004).
4. Kaartinen, V. & Warburton, D. Nat. Genet. 33,
331332 (2003).
5. Derynck, R., Akhurst, R.J. & Balmain, A. Nat. Genet.
29, 117129 (2001).
6. Bhowmick, N.A. et al. Science 303, 848851
7. Abdalla, S.A. et al. Eur. J. Resp. 23, 373377
(2004).
8. Hasham, S.N. et al. Circulation 107, 31843190
(2003).
9. Tang, Y. et al. Hum. Mol. Genet. 12, 15791589
(2003).
10. Cambien, F. et al. Hypertension 28, 881887 (1996).
11. Holweg, C.T. et al. J. Heart Lung Transplant. 20,
12. Beranek, M. et al. Am. J. Med. Genet. 109, 278283
(2002).
13. Sugiura, Y. et al. Ann. Rheum. Dis. 61, 826828
(2002).
14. Crilly, A., Hamilton, J., Clark, C.J., Jardine, A. &
Madhok, R. Ann. Rheum. Dis. 61, 678681 (2002).
15. Gewaltig, J. et al. Clin. Chim. Acta 316, 8394
(2002).

Experiencing VEGF
2004 Nature Publishing Group http://www.nature.com/naturegenetics

David A Greenberg & Kunlin Jin

Vascular endothelial growth factor (VEGF) stimulates angiogenesis, protects neurons from ischemic and degenerative
disorders, and triggers the birth of new nerve cells in the adult brain. Now we discover that it is also important for
memory and learning.

In late eighteenth-century London, the
Scottish anatomist and surgeon John Hunter
interrupted the arterial supply to a stags
antler and found that blood flow was reestab-
lished through newly formed vessels. Such
vascular sprouting, or angiogenesis, is crucial
to the survival of developing tissues and neo-
plasms as they outgrow their blood supply,
and measures to stimulate or inhibit angio-
genesis are prospective treatments for
ischemia and cancer. Angiogenesis is com-
plex, but its most celebrated mediator is vas-
cular endothelial growth factor, or VEGF, a
hypoxia-inducible protein that promotes the
proliferation and survival of vascular
endothelial cells. VEGF and its tyrosine
kinase receptors, VEGFR1 (also called Flt1)
and VEGFR2 (also called KDR or Flk1), are
also expressed by nonvascular cells including
neurons, suggesting that they have a broader
physiological role.
Studies on the neuronal effects of VEGF
establish at least three types of action. First,
VEGF exerts neurotrophic effects manifested
by increased axonal outgrowth and improved
cell survival in neuronal cultures1,2. Second,
VEGF protects neurons from insults such as
hypoxia3 and glutamate toxicity4, which may
explain its ability to reduce injury and improve
outcome after experimental stroke57; con-
VEGF and the ventricle of memory
On page 827 of this issue, Matthew During
and colleagues provide evidence linking
VEGF to a form of neurogenesis that occurs
as a result of experience and may be involved
in memory and learning (Fig. 1).
Environmental enrichment, maze training
and physical exercise are known to stimulate
hippocampal neurogenesis, but the underly-
ing mechanisms are unclear. To explore this
connection, the authors probed rat hip-
pocampus for mRNAs induced by environ-
mental enrichment and by training in a
Morris water maze. They found that VEGF
a b
Morris
water maze
Dentate
gyrus
VEGF
expression increased in both of these settings,
but not after physical exercise alone.
Intracerebral injection of an adeno-associated
virus vector expressing human VEGF
increased hippocampal VEGF expression and
enhanced performance in tests of associative
(passive avoidance) and spatial (Morris water
maze) learning, both of which depend on
hippocampal function. Placental growth fac-
tor (PGF), a VEGF family member that inter-
acts with VEGFR1 but not VEGFR2, had no
such effect, despite the fact that, like VEGF, it
stimulated angiogenesis. Therefore, neither
VEGFR1 activation nor angiogenesis can
Michael Pace

versely, defective VEGF signaling may con-

tribute to motor neuron disease811. Third,
VEGF stimulates neurogenesis, or the produc-
tion of new neurons, in the adult songbird12
and rodent13 brain.
David A. Greenberg and Kunlin Jin are at the
Buck Institute for Age Research, Novato,
California, USA.
e-mail: dgreenberg@buckinstitute.org
Neuron precursor
Figure 1 VEGF links experience to neurogenesis. (a) Water maze training (or other forms of learning or
environmental enrichment) operates through an unknown mechanism to increase VEGF expression in
the hippocampus. VEGF interacts with VEGFR2 receptors on neuronal precursor cells, or on an
unidentified intervening (e.g., endothelial) cell, to stimulate neurogenesis in the hippocampal dentate
gyrus. (b) When stimulated by VEGF, neuronal precursor cells (yellow) in the dentate gyrus subgranular
zone migrate into the dentate granule cell layer, where they differentiate into mature neurons (red).

792 VOLUME 36 | NUMBER 8 | AUGUST 2004 NATURE GENETICS

 CORRIGENDA

Corrigendum: TGF signaling in health and disease

R J Akhurst
Nat. Genet. 36, 790 (2004).

On page 791, in the second column, the second full sentence should begin, Camurati-Engelmann syndrome, a bone overgrowth defect due to
TGFB1 missense mutations....

Corrigendum: A functional variant of SUMO4, a new IB modifier, is

2004 Nature Publishing Group http://www.nature.com/naturegenetics

associated with type 1 diabetes

D Guo, M Li, Y Zhang, P Yang, S Eckenrode, D Hopkins, W Zheng, S Purohit, R H Podolsky, A Muir, J Wang, Z Dong, T Brusko,
M Atkinson, P Pozzilli, A Zeidler, L J Raffel, C O Jacob, Y Park, M Serrano-Rios, M T Martinez Larrad, Z Zhang, H-J Garchon, J-F Bach,
J I Rotter, J-X She & C-Y Wang
Nat. Genet. 36, 837841 (2004).

On page 837, in the right column, the first full sentence should begin, The G variant of this SNP had a higher frequency in affected individuals
from the US (62.7%) than in matched controls (52.4%; P = 0.0008)...
In Figure 2, the arrow indicating the direction of transcription of SUMO4 should be pointing to the right.
The version of Supplementary Table 1 that initially appeared online was incorrect and has been replaced with the correct version.

Corrigendum: Mutations in RDH12 encoding a photoreceptor cell retinol

dehydrogenase cause severe childhood-onset retinal dystrophy
A R Janecke, D A Thompson, G Utermann, C Becker, C A Hbner, E Schmid, C L McHenry, A R Nair, F Rschendorf, J Heckenlively,
B Wissinger, P Nrnberg & A Gal
Nat. Genet. 36, 850854 (2004).

In Figure 2c, 805delCCCTG should read 806delCCCTG. In Figure 3, the y axis in panel a should read Absorbance 325 nm and the y axis in
panel c should read AT-ral peak area (A368nm).

1024 VOLUME 36 | NUMBER 9 | SEPTEMBER 2004 NATURE GENETICS