First published online 2 December 2004
doi: 10.1242/dev.01571
Development 132, 177-187 (2005)
Published by The Company of Biologists 2005
Chx10 repression of Mitf is required for the maintenance of mammalian neuroretinal identity
D. Jonathan Horsford1,2,3,
Minh-Thanh T. Nguyen4,*,
Grant C. Sellar5,
,
Rashmi Kothary6,7,
Heinz Arnheiter4 and
Roderick R. McInnes1,2,3,8,
1 Program in Developmental Biology, The Research Institute, Hospital for Sick
Children, Toronto, Ontario M5G 1X8, Canada
2 Program in Genetics, The Research Institute, Hospital for Sick Children,
Toronto, Ontario M5G 1X8, Canada
3 Department of Molecular and Medical Genetics, University of Toronto, Toronto,
Ontario, Canada
4 Laboratory of Developmental Neurogenetics, National Institute of Neurological
Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892,
USA
5 MRC Human Genetics Unit, Western General Hospital, Crewe Road South, Edinburgh
EH4 2XU, UK
6 Ottawa Health Research Institute, Ottawa, Ontario K1H 8L6, Canada
7 Department of Cellular and Molecular Medicine, and University of Ottawa Center
for Neuromuscular Disease, University of Ottawa, Ottawa, Ontario K1H 8M5,
Canada
8 Department of Pediatrics, University of Toronto, Toronto, Ontario,
Canada

View larger version (94K):
[in a new window]
|
Fig. 1. Mitf is ectopically expressed in the Chx10or-J/or-J NR.
(A) In the eye of an E10.5 wild-type (+/+) embryo, Mitf mRNA is
present only in the presumptive RPE (arrow). (B) In an E10.5
Chx10or-J/or-J (J/J) mutant eye, Mitf is
expressed normally in the presumptive RPE (arrow), and ectopically in the NR
(arrowhead). (C) By E13.5, Mitf is localized to the RPE and
presumptive ciliary margin (arrow) in a wild-type embryo. (D) E13.5
Chx10or-J/or-J animals continue to misexpress
Mitf in the NR (arrowhead). In addition, the presumptive ciliary
margin domain (arrow) is expanded, as evidenced by a larger area of increased
Mitf expression. (E) Mitf protein levels mirror the RPE and
presumptive ciliary margin (arrow) Mitf mRNA expression domains in a
wild-type E13.5 embryo. (F) Mitf protein is synthesized in the E13.5
Chx10or-J/or-J NR (arrowhead) and the enlarged presumptive
ciliary margin region can also be clearly identified (arrow). Scale bars: 25
µm in A,B,D; 32 µm in C; 50 µm in E,F.
|
|

View larger version (97K):
[in a new window]
|
Fig. 2. Tyr, Tyrp1 and Dct expression in the cilary margin is
expanded in the Chx10or-J/or-J E13.5 NR. (A) Tyr
is expressed in a wild-type E13.5 RPE and ciliary margin (arrow). (B) By
contrast, in the Chx10or-J/or-J E13.5 ciliary margin, the
Tyr expression domain is expanded (arrow). (C) Tyrp1 is
expressed in the wild-type RPE and ciliary margin (arrow). (D) Tyrp1
expression in the Chx10or-J/or-J ciliary margin is
expanded (arrow). (E) A wild-type section showing expression of Dct
in the RPE and ciliary margin (arrow). (F) In the
Chx10or-J/or-J ciliary margin, the Dct expression
domain is enlarged (arrow). Scale bars: 40 µm in A; 32 µm in B,C,E; 25
µm in D,F.
|
|

View larger version (74K):
[in a new window]
|
Fig. 3 . Chx10 and Mitf are necessary for their reciprocal
Mitf and Chx10 mutant retinal phenotypes. (A) A Hematoxylin
and Eosin-stained coronal section through a wild-type (+/+; +/+) P0
eye showing normal morphology. (B) An identical Hematoxylin and Eosin-stained
section through an Mitf wild-type
(+/+);Chx10or-J/or-J (J/J) eye showing the thin,
hypocellular, non-laminated NR (arrowhead). (C) The loss of one copy of
Mitf (mi/+) on the Chx10 mutant background
(J/J) results in a thicker, multicellular, laminated NR (arrowhead).
Neuroretinal rescue in the mi/+;J/J eye (n=3/3). (D) Double
Mitf;Chx10 mutants (mi/mi;J/J) have a dramatic normalization
of the NR (arrowhead) when compared with the
Chx10or-J/or-J mouse (B). Neuroretinal rescue in the
mi/mi;J/J eye (n=4/4). (D') An enlargement of the NR
(shown in the region of the arrowhead in D) illustrating the lamination of the
double mutant NR. nbl, neuroblastic layer; ipl, inner plexiform layer; gcl,
ganglion cell layer. (E) Mitfmi/mi (mi/mi) mice
lacking one copy of Chx10 (J/+) express the neuroretinal
marker protein Pax6 in a thickened neuroretinal-like layer (NRLL) in the
dorsal part of the RPE (arrow) and in the NR (arrowhead). (F) The loss of both
copies of Chx10 (J/J) in the Mitf mutant background
(mi/mi) results in a normalization of the thickness of the NRLL in
the dorsal RPE [highlighted by the expression of Pax6 (arrow)]. RPE rescue in
mice with a background of more than 90% 129/SvJ (n=2/4). (G)
In a second genetic background (see below), the Mitf mutation
(mi/mi) results in a more dramatic NRLL in an animal heterozygous for
the Chx10or-J mutation (J/+). Pax6
expression is seen in both the NR (arrowhead) and NRLL (arrow). (H) Double
Mitf;Chx10 mutants (mi/mi;J/J) have a normalized RPE
phenotype, highlighted by the loss of ectopic Pax6-expressing tissue
in the RPE (arrow). RPE rescue in a mixed 129/SvJ;B6C3He background
(n=2/2). Scale bars: 80 µm in A-D; 20 µm in D'; 25 µm
in E,F; 31 µm in G,H.
|
|

View larger version (70K):
[in a new window]
|
Fig. 4. Mitf and Chx10 function together in a dose-dependent
antagonistic fashion to regulate retinal cell identity. (A) An unstained
control Chx10or-J/or-J retinal section at P0 (labeled
J/J above the column of panels) shows pigmentation only in the RPE
monolayer (arrow) and in the ciliary margin (square bracket). The NR
(arrowhead) is unpigmented. (B) An enlargement of (A) displays the pigmented
RPE (arrow) and non-pigmented NR (arrowhead). The broken white line indicates
the edge of the NR. (C) NR-MITF/+;Chx10or-J/or-J
littermates have a normal RPE (arrow), greatly expanded ciliary margin (square
bracket) and a pigmented monolayer (PML) instead of a NR (arrowhead). (D) An
enlargement of (C) shows the pigmented RPE (arrow) and the PML (arrowhead).
(E) The Chx10or-J/or-J NR (arrowhead) expresses the
neural-specific cell-adhesion molecule NCAM, while the RPE does not express
NCAM (arrow). (F) The PML also expresses NCAM (arrowhead); the RPE does not
(arrow). (G) The Chx10or-J/or-J NR expresses Pax6
(arrowhead), while the RPE does not (arrow). (H) The nuclei of the PML
contains the neuroretinal marker Pax6 (arrowhead) in contrast to the RPE,
which lacks Pax6 (arrow). Pax6 subcellular localization changes compared with
that in a Chx10or-J/or-J NR (G), which we verified with a
second independent Pax6 antibody (data not shown). The mechanism mediating the
differential localization is unknown, and its significance is unclear. (I) In
situ hybridization of a Chx10or-J/or-J eye at E11.5 shows
normal neuroretinal Rax expression (arrowhead). (J) A
NR-MITF/+;Chx10or-J/or-J littermate also expresses
Rax in the NR. Normal Rax expression (arrowhead) at E11.5 in
NR-MITF-1, n=5/5. (K) A Chx10or-J/or-J animal
expresses Dct in the RPE (arrow) and presumptive ciliary margin
(asterisk). (L) An identical Dct expression pattern is seen in
NR-MITF/+;Chx10or-J/or-J littermates. Normal Dct
or Tyr (not shown) expression at E11.5 in NR-MITF-1, n=5/5.
(M) An unstained P0 coronal section of an Mitfmi/+ eye
(labeled mi/+) showing RPE pigmentation. The pigmentation present in
the NR (asterisk) is an artifact of the dissection; it is RPE tissue that has
adhered to the NR. (N) RPE-CHX10/+;Mitfmi/+ littermates
have a dramatic decrease in pigmentation in the RPE (arrow). (O) Hematoxylin
and Eosin-stained P0 coronal section of an Mitfmi/+ eye
has a normal RPE (arrow). (P) A higher magnification view of (O) shows the
RPE. (Q) RPE-CHX10/+;Mitfmi/+ littermates have a
morphological change in the dorsal RPE; the RPE monolayer has become a
thickened, multicellular structure (arrow). (R) An enlargement of Q
illustrates the thickened multicellular dorsal RPE (arrow). (S) The
Mitfmi/+ NR expresses the homeodomain protein Pax6
(arrowhead), while the RPE expresses very low levels of Pax6 (arrow). (T) Pax6
is present in the thickened RPE (arrow) as well as the NR (arrowhead) in
RPE-CHX10/+;Mitfmi/+ mice, suggesting that this RPE
structure is a NRLL. (U) Both the NR (arrowhead) and the NRLL (arrow) in
RPE-CHX10/+;Mitfmi/+ mice express Rax mRNA. (V)
The NRLL in RPE-CHX10/+;Mitfmi/+ mice expresses mouse
Chx10 (arrow), as does the NR (arrowhead). (W) Graphical
representation of the percentage frequency of the PML phenotype in different
NR MITF/+;Chx10or-J/or-J transgenic lines. The numbers
within the bars indicate the number of individuals examined for each
transgenic line. NR-MITF-1, n=14/19; NR-MITF-2, n=4/4. (X)
Graphical representation of the percent frequency of changes in RPE phenotype
in the RPE-CHX10/+;Mitfmi/+ transgenic lines. The
depigmentation and NRLL phenotypes are shown. Numbers within the bars indicate
the number of individuals tested. RPE-CHX10-1, depigmentation,
n=6/13; four animals with no pigment were tested for the presence of
an NRLL: n=2/4 (denoted by the asterisk). RPE-CHX10-2, both
depigmentation and NRLL phenotypes: n=1/5. Scale bars: 50 µm in
A,C; 25 µm in B,D-L,S,T; 80 µm in M,N,O,Q; 12.5 µm in P,R,U,V.
|
|

View larger version (163K):
[in a new window]
|
Fig. 5. Chx10 is necessary to maintain neuroretinal cell identity. (A) The
majority (n=18/24) of Chx10or-J/or-J
(J/J) individuals from a mixed genetic background
(129/SvJ;C57BL/6) have a `salt-and-pepper' neuroretinal phenotype
consisting of pigmented and non-pigmented cells (arrowhead) and a normal RPE
(arrow). (B) A higher magnification view of A clearly shows the cellular
pigment phenotype in the NR (arrowhead). The broken white lines indicate the
edges of the NR. For comparison, a 129/SvJ Chx10or-J/or-J
non-pigmented NR is shown in Fig.
4A,B. (C) A subset of Chx10or-J/or-J mice
(n=2/24) from a mixed genetic background have a PML (arrowhead)
instead of a NR and a normal RPE (arrow). (D) An enlargement of (C) focusing
on the PML (arrowhead). Scale bars: 50 µm in A,C; 12.5 µm in B,D.
|
|

View larger version (63K):
[in a new window]
|
Fig. 6. Chx10 lies downstream of FGF in mouse neuroretinal cell identity
decisions. (A) An OV culture from a wild-type animal (+/+) dissected at E9-9.5
and grown for 3 days in the presence of a bovine serum albumin-coated bead
(BSA) results in normal development of the NR (arrowhead) and RPE (arrow),
including pigmentation and expression of Mitf protein (green). (B) By
contrast, a culture of wild-type OV in the presence of an FGF2-coated bead
results in a change in cell identity from an RPE to a NR (arrow), as evidenced
by a loss of pigmentation and greatly reduced Mitf expression, although the NR
appears normal (arrowhead). (C) A Chx10or-J/or-J OV
cultured with an FGF2-coated bead has no effect on the identity of the RPE, as
shown by the pigmentation of the RPE and the expression of Mitf (arrow). The
ectopic Mitf is clearly apparent in the Chx10or-J/or-J NR
(arrowhead). (D) Graphical representation of the percent frequency of normal
RPE pigmentation in optic vesicle cultures incubated with beads coated in BSA,
FGF1 or FGF2. Genotype is indicated by bar color (black, wild-type; grey,
Chx10or-J/+; white, Chx10or-J/or-J).
The asterisks represent experiments that were not performed, rather than a
value of zero. The numbers above the bars indicate the number of individuals
tested. RPE pigmentation: +/+ BSA, 10/10; +/+ FGF2, 1/8; J/J FGF2,
23/29; J/J BSA, 10/10; J/J FGF1, 9/10; J/+ FGF2,
4/28. Scale bars: 20 µm in A-C.
|
|

View larger version (25K):
[in a new window]
|
Fig. 7. A model of the specification, organization and maintenance of vertebrate
retinal cells. (A) Retinal cell type specification occurs prior to the OV
stage. By the OV stage depicted here, retinal cells are already biased to
become either RPE or neuroretinal cells, as shown by the striped green and
yellow region. (B) When the surface ectoderm comes in close contact with the
OV, FGF1 and/or FGF2 from the surface ectoderm signals through Chx10 to
organize the NR, adjacent to the future lens. Chx10 then represses the
expression of Mitf (directly or indirectly) to maintain neuroretinal
cell identity, perhaps through the regulation of Fgf8, Fgf9 and/or
Fgf15 and other neuroretinal genes. Cells that are distant from the
surface ectoderm, and thus from FGF1 and/or FGF2, do not express Chx10,
allowing Mitf expression to continue, thus organizing the RPE at the
back of the developing eye. The RPE, by contrast, appears to be organized by
activin signals from the posterior ocular mesenchyme
(Fuhrmann et al., 2000 ),
signals that may act through Mitf and other genes essential for RPE
formation, such as Otx1, Otx2
(Martinez-Morales et al.,
2001 ), Pax2 and Pax6
(Baumer et al., 2003 ). Mitf
appears to maintain RPE cell identity by the activation of downstream genes,
such as pigmentation enzyme-encoding genes. Finally, although the mechanism is
unknown, Mitf may also negatively regulate Chx10 to maintain RPE cell
identity.
|
|

CiteULike
Complore
Connotea
Del.icio.us
Digg
Reddit
Technorati
Twitter What's this?
© The Company of Biologists Ltd 2005