Mutational probing of the forkhead domain of the transcription factor FOXL2 provides insights into the pathogenicity of naturally occurring mutations.

Abstract

Mutations of the transcription factor FOXL2, involved in cranio-facial and ovarian development, lead to the Blepharophimosis Syndrome. Here, we have systematically replaced the amino acids of the helices of the forkhead domain (FHD) of FOXL2 by glycine residues to assess the impact of such substitutions. A number of mutations lead to protein mislocalization, aggregation and to partial or complete loss of transactivation ability on a series of luciferase reporter systems. To rationalize the results of this glycine mutation scan, we have modeled the structure of the FHD by comparison with crystallographic data available for other FHDs. We failed to detect a clear-cut correlation between protein mislocalization or aggregation and the position of the mutation. However, we found that the localization of the side chain of each amino acid strongly correlated with the impact of its mutation on FOXL2 transactivation capacity. Indeed, when the side chains of the amino acids involved in the helices of the forkhead are supposed to point towards the hydrophobic core formed by the three main helices, a loss of function was observed. On the contrary, if the side chains point outward the hydrophobic core, protein function was preserved. The extension of this analysis to natural mutants shows that a similar correlation can be found for BPES mutations associated or not with ovarian dysfunction. Our findings reveal new insights into the molecular effects of FOXL2 mutations affecting the FHD, which represent two-thirds of intragenic mutations, and provide the first predictive tool of their effects.

Full text

Page 1

Mutational probing of the forkhead domain of the
transcription factor FOXL2 provides insights into
the pathogenicity of naturally occurring mutations
Anne-Laure Todeschini1,2, Aure ´lie Dipietromaria1,2, David L’Ho ˆte1,2, Fatima Zohra Boucham1,
Adrien B. Georges1,2,3, P.J. Eswari Pandaranayaka5, Sankaran Krishnaswamy5, Isabelle Rivals4,
Claude Bazin1,2and Reiner A. Veitia1,2,∗
1CNRS UMR 7592, Institut Jacques Monod, Equipe Ge ´ne ´tique et Ge ´nomique du De ´veloppement Gonadique, 75205
Paris Cedex 13, France,2Universite ´ Paris Diderot-Paris VII, 75205 Paris Cedex 13, France,3Ecole Normale
Supe ´rieure de Paris, 75005 Paris, France,4Equipe de Statistique Applique ´e, ESPCI Paris Tech, 75005 Paris, France
and5School of Biotechnology, Madurai Kamaraj University, 625021 Madurai, India
Received May 3, 2011; Revised and Accepted May 26, 2011
Mutations of the transcription factor FOXL2, involved in cranio-facial and ovarian development, lead to the
Blepharophimosis Syndrome. Here, we have systematically replaced the amino acids of the helices of the
forkhead domain (FHD) of FOXL2 by glycine residues to assess the impact of such substitutions.
A number of mutations lead to protein mislocalization, aggregation and to partial or complete loss of trans-
activation ability on a series of luciferase reporter systems. To rationalize the results of this glycine mutation
scan, we have modeled the structure of the FHD by comparison with crystallographic data available for other
FHDs. We failed to detect a clear-cut correlation between protein mislocalization or aggregation and the pos-
ition of the mutation. However, we found that the localization of the side chain of each amino acid strongly
correlated with the impact of its mutation on FOXL2 transactivation capacity. Indeed, when the side chains of
the amino acids involved in the helices of the forkhead are supposed to point towards the hydrophobic core
formed by the three main helices, a loss of function was observed. On the contrary, if the side chains point
outward the hydrophobic core, protein function was preserved. The extension of this analysis to natural
mutants shows that a similar correlation can be found for BPES mutations associated or not with ovarian
dysfunction. Our findings reveal new insights into the molecular effects of FOXL2 mutations affecting the
FHD, which represent two-thirds of intragenic mutations, and provide the first predictive tool of their effects.
INTRODUCTION
The Blepharophimosis Ptosis Epicanthus Inversus Syndrome
(BPES, MIM 110100) is a genetic disease leading to cranio-
facial malformations associated with premature ovarian
failure (POF) (type I BPES) or occurring alone (type II
BPES) (1). The FOXL2 gene, encoding a transcription
factor, is mutated in patients with BPES (2). The protein
sequence of FOXL2 is highly conserved and contains a
characteristic DNA-binding domain called forkhead (FHD)
(3). The FHD is also conserved among paralogs (FOX pro-
teins) across a wide evolutionary span (4). It folds into a
variant of the helix-turn-helix motif and is mainly composed
of three a-helices (H1, H2 and H3) and two large loops or
‘wings’ (W1 and W2) (5). An additional a-helix (H4) can
be found between H2 and H3 in some FHDs. The strong
primary sequence conservation of this domain implies that
its 3D-structure is essential for protein function. FOXL2 also
carries a polyAlanine tract of 14 residues strictly conserved
among eutherian mammals, the functional relevance of
which, if any at all, is not known yet (3,6). FOXL2 is a
nuclear protein present in peri-occular and ovarian follicular
cells, among other expression sites (3,7). Targeted deletion
of Foxl2 in mouse induces an accelerated depletion of the
∗To whom correspondence should be addressed at: Institut Jacques Monod, Ba ˆtiment Buffon/Pie `ce 555B, 15 rue He ´le `ne Brion, 75205 Paris Cedex 13,
France. Email: veitia.reiner@ijm.univ-paris-diderot.fr
# The Author 2011. Published by Oxford University Press. All rights reserved.
For Permissions, please email: journals.permissions@oup.com
Human Molecular Genetics, 2011, Vol. 20, No. 17
doi:10.1093/hmg/ddr244
Advance Access published on June 1, 2011
3376–3385
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 2

ovarian follicular pool that confirms its critical role for a
proper differentiation of granulosa cells during folliculogen-
esis (8,9). Several analyses have implicated FOXL2 in the
regulation of steroid metabolism, cell cycle, reactive oxygen
species detoxification, apoptosis and inflammation (10–13).
A large variety of FOXL2 mutations has been described thus
far (details at http://medgen.ugent.be/foxl2). More than
two-thirds of BPES patients carry intragenic FOXL2 mutations
(14,15) and expansions of the polyAlanine tract of the protein
from 14 to 24 residues account for one-third of mutations in
the coding region (14,15). We have shown that such
polyAlanine expansions lead to the formation of intranuclear
aggregates and to a mislocalization of the protein to the cyto-
plasm (16,17). We have previously studied the molecular con-
sequences of a series of naturally occurring FOXL2 missense
mutations affecting the FHD on subcellular localization and
transactivation (18). The mutant proteins could be grouped
into four classes: (i) those displaying a normal nuclear distri-
bution (H104N, N109K); (ii) those displaying nuclear aggre-
gation (E69K, S101R); (iii) those inducing nuclear and
cytoplasmic aggregation (S58L, I63T, A66V, I80T, I84N,
F90S, W98G, I102T, R103C, L106F, L106P) and (iv) those
with cytoplasmicaggregation
mutations lead to a classical BPES phenotype, but the predic-
tive value of alterations of subcellular localization for geno-
type–phenotypecorrelations
aggregation and mislocalization of mutated FOXL2 tended
to impair its transactivation ability. More recently, using two
luciferase-based promoter reporter assays, we were able to
find a direct correlation between the transcriptional activity
of FOXL2 variants and the BPES type, enabling us to
predict the risk of POF associated with these mutations
(which is otherwise a difficult task) (19). In this previous
work, we also detected a loose correlation between the pro-
portion of cells displaying subcellular mislocalization and
aggregation of mutated FOXL2 and the functional activity
on the reporter promoters. Recently, Shah et al. (20) identified
the recurring somatic mutation c.402C . G in the sequence of
FOXL2 leading to the amino acid substitution p.C134W in
ovarian granulosa cell tumors (OGCTs). The presence of
this mutation has been confirmed in .95% of adult-type
OGCTs by different laboratories (21,22). This mutation
maps to the second wing of the FHD. Shah et al. (20) and our-
selves (22) failed to detect any mislocalization of the mutated
protein. More recently, it has been shown that the wild-type
(WT) and mutant FOXL2 display differential pro-apoptotic
activities. Specifically, WT FOXL2 induces significant granu-
losa cell death while the mutant version exhibited a minimal
activity (23).
In the present study, we set out to assess, in a systematic
way, the impact of artificial mutations in the helices of the
FHD of FOXL2 on its intracellular physical state (i.e. aggrega-
tion, mislocalization) and activity. For this, we have generated
(S70I). Thesemissense
islimited.However,
20 mutated versions of FOXL2 in which we have basically
replaced one residue every two positions in the a-helices of
the FHD by a glycine. We have named the whole process a
‘Glycine mutation scan’ of the FHD helices. We have
chosen glycine because it tends to disrupt a-helices owing to
its high conformational flexibility (24). Indeed, a feature of
the C-terminus of helices is the frequent presence of a
glycine (i.e. the C-cap) in a left-handed a conformation with
the NH making a hydrogen bond to the CO of amino acid
n-5 (25). Moreover, in the context of a protein, glycine does
not project any hydrophobic lateral chain that might help
stabilize the structure of the domain by promoting interactions
with a hydrophobic core (as might be the case of Alanine,
which was another candidate, with the highest helix-forming
propensity). Withstanding a glycine mutation within the
a-helices would also show how the tertiary structure is resili-
ent (or sensitive) to such helix-disrupting mutations. We
hypothesized that such a systematic analysis would provide
a framework to understand the pathogenic mechanism of
natural mutations.
Here, we find that the predicted spatial orientation of the
side chains of the mutated residues with respect to the hydro-
phobic core formed by the three main helices of the FHD is the
key parameter to explain the functional effects of the
mutations. On the one hand, artificial mutations of residues
whose lateral chains point outward the hydrophobic core are
well tolerated, leading to normal activity in a wide series of
luciferase reporter assays. On the other hand, mutations of
residues pointing inward impair FOXL2 activity. Interestingly,
natural mutations follow the same pattern. The side chains of
the amino acids whose substitutions are involved in type I
BPES (i.e. the more damaging ones) tend to point towards
the hydrophobic core, and vice-versa for the type II-inducing
mutations. Taken altogether, our results show that a pertur-
bation of the hydrophobic core stabilizing the helices of the
FHD leads to protein malfunctioning and pathology.
RESULTS AND DISCUSSION
Mutations in the alpha-helices of the FHD of FOXL2 lead
to variable protein mislocalization and aggregation
In order to better understand the relevance of the a-helices to
maintain the structure of a FHD, we have constructed a series
of mutants in the context of the transcription factor FOXL2.
Specifically, we have basically replaced one amino acid
every two positions by a glycine residue in the helices of the
FHD (Fig. 1). Our 20 constructs express a mutated FOXL2
version in fusion with the green fluorescent protein (GFP,
Primers sequences in Supplementary Material, Table S1).
We have started our exploration by assessing the impact of
the glycine mutation scan on the localization and physical
state (i.e. aggregation or not) of the mutant proteins. For
Figure 1. Sequence of FOXL2 forkhead from Pro49 to Cys115. Secondary structure elements are indicated at the bottom of the diagram. Color code of the amino
acid residues: grey ¼ WT sequence; black ¼ residue mutated in this work (glycine scan); underlined ¼ natural FOXL2 missense mutation used in this work.
Human Molecular Genetics, 2011, Vol. 20, No. 173377
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 3

this, we transiently transfected the mutated constructs in
COS-7 cells to evaluate the proportion of cells displaying
cytoplasmic staining and/or aggregation as well as intranuclear
aggregation, by fluorescence microscopy (18). As expected,
the WT protein (FOXL2-WT) displayed a proper nuclear
localization, (Fig. 2). The same pattern was also observed
for F90G, K93G, N94G, W98G, S101G and H104G
mutants. However, for the other mutants, we observed granu-
lar or massive nuclear aggregates. For instance, mutants
Y59G, K87G and L108G presented 10% of nuclear aggrega-
tion. This proportion rose to 20% for the mutated versions
I63G, M65G, L77G, Y81G, N100G and I102G and to more
than 30% for Y83G, I85G, Y91G and L106G. With the excep-
tion of L77G, Y91G, I102G and L106G, which displayed
cytoplasmic aggregates, the rest of the mutants were localized
basically in the nucleus. We failed to detect any obvious
correlation between nuclear and/or cytoplasmic aggregation
and mislocalization and the position of the mutation in our
series of 20 variants (Fig. 2). Furthermore, we failed to find
any clear link between mutations in a particular helix and a
tendency to induce aggregation or mislocalization. This con-
trasts with previous data on other forkhead factors. Naturally
occurring missense mutations in the FHDs of FOXC1,
FOXC2 and FOXP2 have been shown to lead to defective
protein localization loosely correlating with impaired transac-
tivation (26–29). A previous predictive model proposed that
mutations in the first helix of the FHD might destabilize
protein–DNA interactions, interfering with transactivation
and inducing subcellular mislocalization (28). Mutations in
the second helix would not affect subcellular localization,
but would decrease transcriptional activity. Finally, mutations
in the DNA-recognition helix (the third one) would impact
Figure 2. Subcellular localization of wild-type and glycine scan mutant FOXL2 proteins fused to the GFP. The middle panel corresponds to the most
representative subcellular localization of each mutant or the WT protein in transiently transfected COS-7 cells. The left panel corresponds to nuclei stained
with Hoescht 33342. The right panel is a merge of both. The graphs display the quantitative representation of subcellular localization of FOXL2 variants. Per-
centages were calculated over 500 transfected COS-7 cells.
3378Human Molecular Genetics, 2011, Vol. 20, No. 17
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 4

nuclear localization and DNA binding. The divergence
between these observations and our results might reside on
specific differences of the relevant proteins (i.e. FOXL2
versus FOXC1 and 2), on the difference in the nature of the
amino acid substitutions or on the number of mutations
studied. Our present and previous data (involving natural
mutations) show that the predictive model for missense
mutations in the FHD of FOXC1 and FOXC2 do not apply
in a straightforward manner to FOXL2 (19).
Mutations in the helices of the FHD of FOXL2 lead
to variable transactivation impairment on a series
of luciferase reporter systems
Next, we assessed whether the mutated variants were still able
to transactivate FOXL2 target promoters in the KGN human
granulosa-cell tumor cell line. KGN cells naturally express
FOXL2 and constitute a suitable cellular model to study its
activity. Specifically, we used 12 reporter constructs that
should allow us to detect either intrinsic variations of
FOXL2 transcriptional activity (i.e. promoters containing con-
catemerized FOXL2 response elements) and/or variations due
to potential perturbations of interactions with proteins partners
(naturally occurring promoters). The reporters used in this
study are listed in Table 1. KGN cells were co-transfected
with the promoter-luciferase construct and one of the con-
structs driving the expression of a FOXL2 variant (along
with the plasmid pRL-RSV to correct for differences in trans-
fection levels) (Fig. 3 and Supplementary Material, Table S2).
Figure 3 shows typical results of two reporter promoters
(pSIRT1-luc and 2x-A-FLRE-luc). On the pSIRT1-luc repor-
ter, FOXL2-WT induced a significant increase of luciferase
activity when compared with the empty pcDNA-GFP vector
(Fig. 3A). Interestingly, some FOXL2 glycine mutants dis-
played WT levels of transactivation (M65G, Y83G, I85G,
F90G, K93G, N94G, N100G, S101G, H104G and L108G)
while others displayed significantly reduced activities (I63G,
L77G, Y81G, Y91G, W98G, I102, S107G). On this reporter,
Y59G, K87G and L106G displayed intermediate activities
(different from both FOXL2-WT and empty vector). Similar
results were obtained with the 2x-A-FLRE-luc reporter
(Fig. 3B), but a few differences appeared between these two
experimental systems. Indeed, on 2x-A-FLRE-luc reporter
more mutants displayed an intermediate activity (i.e. Y59G,
Y83G, K87G, F90G and S101G). This might be explained
by a higher sensitivity of the latter reporter to the available
amounts of functional FOXL2 (which ensures a better dis-
crimination).
In order to obtain an integrated view of the functional impact
of the mutations on our series of 12 reporter promoters, we per-
formedahierarchicalclustering(Fig.4).Hierarchicalclustering
creates aseriesofclusters organizedinatreestructure. Theroot
of the tree consists of a single cluster containing all the
mutations, and each of the leaves contains a single mutation.
Between these two extremes, each level of the hierarchy rep-
resents a particular grouping of the data into disjoint clusters
of mutations. Here, the second-highest level clearly classified
the variants into two homogenous clusters: M65G, Y83G,
I85G, K93G, N94G, N100G, L108G and WT on one side, and
Y59G, I63G, L77G, Y81G, K87G, F90G, Y91G, W98G, I102,
H104G and L106G on the other side. The latter group involved
strongly damaging changes that virtually abolished protein
activity. As shown subsequently, this behavior is linked to the
predicted positions of the relevant mutated amino acids within
the helices of the FHD.
We failed to detect a clear link between mislocalization/
aggregation of the mutated proteins and their functional capa-
bilities. For instance, Y83G and I85G were still able to induce
luciferase expression of tested reporters while displaying
nuclear aggregation in about 30% of transfected cells. This
phenomenon probably reflects a difference in the nature of
the aggregates that in some cases might leave enough avail-
able FOXL2 to activate our set of reporter promoters. This
would not be surprising in the light of the obvious differences
in the aggregation patterns of the series of mutated FOXL2
proteins (discussed subsequently). The analysis of the compo-
sition, the dynamics and ultrastructure of different types of
aggregates deserves future studies.
We next asked whether there could be a potential link
between the chemical nature of the substitution (AA versus
Gly) and its functional impact. To estimate the expected
effect of the mutation, we used the scores of the Point
Accepted Mutation (PAM250) similarity matrix. In such
scoring matrices, the amino acid substitutions favored by
Table 1. List of reporter constructs used in this work
Promoter name ContainingBackboneReference
GRAS-luc
pSIRT1-Luc
DK3-luc
2X-FLRE-luc
2Xm-FLRE-luc
4X-FLRE-luc
4Xm-FLRE-luc
pPROMhFOXL2
m2s-luc
pPTGS2-luc
pSOD-luc
2X-A-FLRE-luc/ dAT29Cns
Three repeats of the mouse GnRH Receptor Activating Sequence
1053 bp of the human SIRT1 promoter
1055 bp of the goat Foxl2 promoter
Two FLRE sequences
Two mutant versions of FLRE sequences
Four FLRE sequences
Four mutant versions of FLRE sequences
3372 bp of the human FOXL2 promoter
Two mutant versions and two wt versions of FLRE sequences
Human promoter
3600 bp of the human MnSOD promoter
Two FLRE AT29C sequences
pTKLuc+
PGL3-basic
PGL3-basic
PGL3-basic
PGL3-basic
PGL3-basic
PGL3-basic
PGL4–12
PGL3-basic
pluc-MCS
PGL3-basic
PGL3-basic
(39)
(11)
(40)
(41)
(41)
(41)
(41)
Present study
(41)
(10)
(42)
Present study
The FLRE sequence is TCAAGG, mutant version of FLRE sequence is TCAATT. FLRE AT29C sequence is ACGGTCAAGGTCACGG.
FLRE, FOXL2 responsive element.
Human Molecular Genetics, 2011, Vol. 20, No. 17 3379
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 5

evolution (i.e. a conservative replacement) are assigned ‘simi-
larity’ scores .0 and non-conservative replacements have
scores ,0. Thus, similarity scores .0 suggest that two
amino acids can replace each other without negative func-
tional consequences and vice-versa. The PAM 250 matrix is
supposed to represent the amino acid replaceability in
protein sequences having diverged by 250 mutations per 100
amino acids positions. This comes down to say that the
scores basically reflect the structural/functional replaceability
of the amino acids with a low impact of the structure of the
genetic code (30). Not surprisingly, the average normalized
activities of the FOXL2 variants (calculated over the 12 repor-
ter promoters) displayed a positive correlation with the
PAM250 scores for the glycine substitutions (Pearson’s corre-
lation coefficient R ¼ 0.6, P , 0.01, Fig. 5). This means that
the higher the interchangeability of the particular residue by
a Gly is, the lower functional interference of the mutation.
As shown in what follows, this was also found when naturally
occurring mutations were taken into account.
Molecular modeling of FOXL2 reveals a correlation
between structural and functional effects
We hypothesized that molecular modeling of FOXL2 structure
might help understand the impact of glycine substitutions in
the helices in maintaining the structure of the FHD. Using
Figure 3. Transcriptional activities of FOXL2 mutants (from the glycine scan and natural mutations). Luciferase assays were performed in KGN cells. Relative
lights units (RLU) correspond to the ratio of the activity of the firefly luciferase reporter over that of the Renilla reporter (internal control of transfection effi-
ciency). Transcriptional activity was tested on two promoters: (A) on the pSIRT1-Luc reporter and (B) on the 2X-A-FLRE-Luc reporter. The white bars show the
basal activity of the promoter in absence of FOXL2 overexpression. The black bars show the transcriptional activity of FOXL2-WT overexpression. According to
a Student’s test, the light grey bars show mutants displayed an activity significantly different from WT and not different from pcDNA (P , 0.001). The dark grey
bars show mutants displayed an activity significantly different from pcDNA but not different from WT (P , 0.001). The hatched black bars show mutants
displayed an activity significantly different from WT and pcDNA.
3380 Human Molecular Genetics, 2011, Vol. 20, No. 17
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 6

crystallographic data from other FHDs as a template, we pre-
dicted in silico the localization of the side chain for each
amino acid position (Fig. 6). We sorted the relevant amino
acid positions into three different classes. If the side-chain
of the amino acid residue points towards the core of the three-
helical bundle of the FHD, the residue was classified as ‘in’
(for pointing inward). On the contrary, when the side-chain
points outward the three-helical core, the residue was classi-
fied as ‘out’. In some instances, the side chains point neither
inward nor outward the core and such residues were classified
as ‘not determined’ (Supplementary Material, Table S3). Not
surprisingly, the average hydrophobicity (according to the
Tanford’s scale) (31) of the inward-pointing amino acid resi-
dues, irrespective of whether they were mutated in our study
or not, was much higher (0.4+1.0) than the outward pointing
residues (20.2+1.0). This clearly shows the importance of
hydrophobic interactions to stabilize the structure of the
three helical bundle of the FHD. Concerning the mutated
positions, Y59G, I63G, L77G, Y81G, K87G, Y91G, W98G,
I102G and L106G were classified as ‘in’, whereas M65G,
K93G, N100G, H104G, S107G and L108G were classified
as ‘out’. Y83G, I85G, F90G, N94G and S101G were classified
as ‘not determined’. Interestingly, we found a strong
correlation between the amino acid position in the predicted
FOXL2 structure and its functional impact. Namely, when
the side chain of the amino acid residue points outward the
hydrophobic core, the glycine substitution is still compatible
with a normal activity (with the exception of S107G). On
the contrary, when the side chain points inwards, the glycine
substitution leads to a loss of activity of the variant. This
clear-cut correlation strongly suggests that our structural
model is correct and prompted us to perform further exper-
imental and statistical analyses involving naturally occurring
mutations.
Towards a classification of BPES-inducing FOXL2
mutations based on a structural model
As mentioned earlier, we have previously studied a series of
16 naturally occurring point mutations in FOXL2. Here, we
studied in-depth nine natural mutations known or predicted
to lead to BPES of either type. Because mislocalization and
aggregation of these mutated proteins were previously
assessed (19), we focused on the analysis of their transcrip-
tional activity on a series of luciferase reporters (Fig. 3 and
Supplementary Material, Table S2). The results obtained
were consistent with what we have previously described,
namely, that known or predicted type II mutants still have
transcriptional activity while known or predicted type I
mutants are devoid of transactivation capacity. This fact
prompted us to include in our statistical analyses all the natu-
rally occurring FHD mutations for which data on transactiva-
tion on pSIRT1-luc and 4xFLRE-luc reporters were available
(19). As expected, the combined analysis of the glycine muta-
tional scan and the natural mutations confirms the correlation
between the functional impact of the amino acids substitution
and thecorresponding PAM250scores(Fig.5B).
Figure 4. Hierarchical clustering of the FOXL2 mutations and FOXL2-WT on
the basis of their transactivation activity on a set of 12 reporter promoters.
Black boxes correspond to an activity similar to the average activity of the
column (reporter promoters), red boxes to higher values and green ones to
lower values. The height of each ‘U’ of the binary tree is proportional to
the distance between the two clusters it joins together.
Figure 5. Positive correlation between the activities of FOXL2 variants and
the scores from the look-up PAM250 matrix for the corresponding amino
acid substitutions. (A) Involves the artificial glycine substitutions and their
normalized activities averaged over 12 reporter promoters. (B) Includes both
glycine substitutions and naturally occurring mutations and their effects on
two reporter promoters (4xFLRE and pSIRT1). The observed positive corre-
lation means that the higher the interchangeability of the particular residue
by a Gly (or another amino acid residue, in the case of natural mutations)
is, the lower functional interference of the mutation.∗∗P , 0.01,∗P , 0.05.
Human Molecular Genetics, 2011, Vol. 20, No. 173381
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 7

Interestingly, the combined analysis of artificial and natural
mutants showed the existence of a loose negative correlation
between the average activity of FOXL2 variants measured
on the pSIRT and 4xFLRE reporters and the % of cells dis-
playing nuclear aggregation (R ¼ 20.35, P , 0.05) and cyto-
plasmic retention (R ¼ 20.39, P , 0.05). We failed to find
such a significant correlation with the artificial mutants
alone, but we have previously reported its existence in the
context of natural mutations (19). We conclude that statistical
significance in the combined analysis involving artificial and
natural mutants is driven by the latter and/or owes to the
increase of the sample size. Whatever the explanation, a nega-
tive correlation between aggregation and mislocalization and
protein activity is not surprising.
When we analyzed in silico the localization of the side
chain for each BPES mutant, we found an almost perfect cor-
relation: for known or predicted type I BPES mutations (i.e.
those abolishing activity on pSIRT and 4xFLRE reporters),
the amino acid side chains of the relevant amino acid positions
point towards the hydrophobic core of the three helical bundle,
whereas for type II (known or predicted) BPES mutations they
point outwards, or at worst cannot be classified. This comes
down to say that strongly damaging naturally occurring
mutations (i.e. type I) are likely to disturb the hydrophobic
interactions expected to stabilize the FHD.
Our results show that both the nature and the position of a
FOXL2 mutation, be it artificial or natural, have a critical
impact on the functional outcome. From a practical perspec-
tive, our structural model provides a reasonable classification
tool of the BPES type for a given mutation occurring in the
helices of the FHD. Indeed, a mutation leading to a pertur-
bation of the hydrophobic core of the FHD would generate a
loss-of-function allele, whose corresponding protein would
be unable to activate even highly sensitive promoters. Such
a mutation would very likely lead to POF. How a single
mutation leads to protein misfolding and aggregation is
Figure 6. Molecular modeling of the helices in the FHD of FOXL2. Each helix (H1–4) is represented as a flat projection using the InsightII software. If the side
chain points ‘in’, amino acid is depicted in pink, if the side chain points ‘out’ or ‘undetermined’ amino acid is depicted in blue.
3382Human Molecular Genetics, 2011, Vol. 20, No. 17
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 8

largely unknown. Thus, it could be interesting to crystallize
the WT and several mutated forms of the FHD of FOXL2 in
order to better understand which amino acids are directly
involved in the formation of the hydrophobic core and
which are the structural consequences of their mutation. It is
also worth noting that interactions with other proteins (or
with DNA?) might play a role in the stabilization of the 3D
fold of the FHD. There are examples of disordered proteins
that adopt folded structures upon binding to their partners.
This is the case for the interactions between domains of the
general transcriptional coactivator CREB-binding protein
(CBP) and the activator for thyroid hormone and retinoid
receptors (ACTRs) (32 and references therein). Interestingly,
both the ACTR and the CBP domains are intrinsically
unfolded in isolation but their co-expression leads to a struc-
tured stoichiometric complex. Accordingly, the unstructured
CBP domain fused to the GFP aggregates in a bacterial
expression system but co-expression of the ACTR domain
improves its folding and solubility. FHD stabilization can
also be influenced by water molecules. Indeed, FOXO–DNA
complexes involve several ordered water molecules, which
could play an important role in their stabilization (reviewed
in 33). It is very likely that a change in these interactions
can prevent a correct transactivation.
In more practical terms, BPES cases bearing mutations that
are predicted to lead to ovarian dysfunction should undergo
frequent monitoring of FSH and AMH levels to identify the
onset of ovarian failure, which can occur between puberty
and 40 years of age (34). Hopefully, new therapies able to
delay the onset of POF will be available in the near future.
MATERIALS AND METHODS
Plasmids and expression vectors
The FOXL2 mutated variants were produced by junction PCR
as previously described (35). A table summarizing the primer
sequences used in this study is available in the online sup-
plement (Supplementary Material, Table S1). Briefly, two
PCR reactions were performed to generate the 5′and 3′por-
tions of the ORF using, respectively, primer pcDNAFOXL2F
and the corresponding mutagenic R primer (Supplementary
Material, Table S1) and pcDNAFOXL2R and the correspond-
ing F primer. After purification of the PCR products, they were
mixed and allowed to undergo five PCR cycles in absence of
primers, to generate the full-length mutated ORF. Then, a final
PCR reaction was performed using primers pcDNAFOXL2F
and pcDNAFOXL2R to further amplify the ORF. The
PCR-amplified ORFs were cloned into the pcDNA3.1/
CT-GFP topoTA cloning vector (Invitrogen, Carlsbad, CA,
USA) to produce recombinant fusion proteins in frame with
the GFP.
Luciferase reporters used in this study are listed and refer-
enced in Table 1. The sequence of the human FOXL2 promo-
ter was amplified by PCR from a BAC sequence containing
the FOXL2locususing
5′-ATGATGGTACCGCTCTATTCTCTCCCGCTCA-3′
pPROMFOXL2-R 5′-ATGATGCTAGCGACAAAGCCGGC
GCGCCGCGG-3′. PCR products were digested with the
NheI and KpnI endonucleases (Boehringer–Mannheim) and
primers pPROMFOXL2-F
and
cloned in pGL4–12 (Promega) to generate luciferase reporter.
All constructs were sequenced to exclude the presence of
PCR-induced mutations.
Cell culture
COS-7 cells (African green monkey), used for protein localiz-
ation studies, were grown and maintained in DMEM medium
supplemented with 10% of fetal bovine serum and 1% of peni-
cillin/streptomycin. The human granulosa-like KGN cells
were used for functional studies (36). They were maintained
in DMEM-F12 medium supplemented with 10% of fetal
bovine serum and 1% of penicillin/streptomycin and seeded
at 4300 cells per well in 96-well culture plates, 24 h before
transfection.
Protein subcellular localization
To assess subcellular localization/aggregation, COS-7 cells
were transfected with FOXL2 constructs, as previously
described (18). These experiments were reproduced in three
independent transfections. Briefly, cells were seeded 24 h
before transfection at 30% of confluence in 24-well plates con-
taining sterile coverslips. Cells were transfected using the
calcium phosphate method (37) with 1 mg of plasmid per
well and rinsed 24 h after transfection. Forty-eight hours
after transfection, cells were washed with phosphate-buffered
saline solution (PBS) and fixed for 15 min with Histofix
(Trend Scientific, Inc.). Cells were washed three times in
PBS, and nuclei were stained with the Hoechst 33342 dye
(Invitrogen, CA, USA). Then, coverslips were mounted on
slides using fluorescence mounting medium DakoCytomaton
(DAKO, CA, USA). Cells were visualized by epifluorescence
microscopy using a Leitz Aristoplan microscope, with a Leitz
objective lens ×100/1.37 Oil with a numerical aperture of
0.17. Image acquisition was done with a JVC KX-F70B
camera. Nuclear and cytoplasmic staining/aggregation were
scored as previously described (18). For each construct, 500
GFP-positive COS-7 cells were analyzed from three different
transfection experiments.
Luciferase reporter assays and hierachical clustering
The transcriptional activity of our set of FOXL2 mutants was
assessed in KGN cells using the Dual-Luciferase Reporter
Assay System (Promega, Madison, WI, USA). As luciferase
reporter, we used a series of previously described or new pro-
moter constructs (Table 1). Each experiment was performed in
six replicates in 96-well culture plates. Cells were transfected
using the calcium phosphate method (37) with 150 ng of
plasmid per well and rinsed 24 h after transfection. Forty-eight
hours after transfection, cells were washed with PBS before
lysis and luciferase activity measurements. To monitor trans-
fection efficiency, a Renilla luciferase vector (pRL-RSV,
Promega, Madison, WI, USA) was co-transfected in all exper-
iments. Luciferase results are reported as relative light units
(RLU, i.e. the ratio of the firefly luciferase and the Renilla
luciferase activities). Statistical significance was estimated
with a Student’s t-test. Error bars represent standard deviation
between replicates.
Human Molecular Genetics, 2011, Vol. 20, No. 173383
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 9

Hierarchical clustering of the mutations on the basis of the
transactivation activity of FOXL2 on the set of reporter pro-
moters was performed using agglomerative clustering with
average linkage (38), using the log 10-transformed average
activities of the luciferase outputs (RLU) after setting the
transactivation value of the empty vector to 1. Prior to cluster-
ing, the missing values in Supplementary Material, Table S2
were estimated using K-nearest neighbors with K ¼ 3 and
the Euclidian distance as metrics.
In silico 3D modeling of FOXL2 FHD mutations
The FHD of FOXL2 was modeled using the FoxP2 FHD
crystal structure (PDB Id: 2A07) as template. Homology
module of Insight 2000 Accelrys software on SGI-O2 was
used for modeling. The mutations were placed within the
FOXL2 model using standard rotamer geometry as in the bio-
polymer module of the Accelrys software. The association of
six molecules of FoxL2 FHD with two DNA molecules, as
seen in the FoxP2 FHD crystal structure, was used to
model the association of the native and mutated FoxL2
FHD, in order to understand the possible role of these
mutants.
SUPPLEMENTARY MATERIAL
Supplementary Material is available at HMG online.
FUNDING
This work was supported by Centre National de la Recherche
Scientifique, La ligue contre le Cancer (Comite ´ de Paris),
l’Universite ´ Paris Diderot-Paris7 and Institut Universitaire de
France.
REFERENCES
1. Zlotogora, J., Sagi, M. and Cohen, T. (1983) The blepharophimosis,
ptosis, and epicanthus inversus syndrome: delineation of two types.
Am. J. Hum. Genet, 35, 1020–1027.
2. Crisponi, L., Deiana, M., Loi, A., Chiappe, F., Uda, M., Amati, P.,
Bisceglia, L., Zelante, L., Nagaraja, R., Porcu, S. et al. (2001) The
putative forkhead transcription factor FOXL2 is mutated in
blepharophimosis/ptosis/epicanthus inversus syndrome. Nat. Genet., 27,
159–166.
3. Cocquet, J., Pailhoux, E., Jaubert, F., Servel, N., Xia, X., Pannetier, M.,
De Baere, E., Messiaen, L., Cotinot, C., Fellous, M. et al. (2002)
Evolution and expression of FOXL2. J. Med. Genet., 39, 916–921.
4. Shimeld, S.M., Degnan, B. and Luke, G.N. (2010) Evolutionary genomics
of the Fox genes: origin of gene families and the ancestry of gene clusters.
Genomics, 95, 256–260.
5. Kaestner, K.H., Knochel, W. and Martinez, D.E. (2000) Unified
nomenclature for the winged helix/forkhead transcription factors. Genes
Dev., 14, 142–146.
6. Cocquet, J., De Baere, E., Gareil, M., Pannetier, M., Xia, X., Fellous, M.
and Veitia, R.A. (2003) Structure, evolution and expression of the FOXL2
transcription unit. Cytogenet. Genome Res., 101, 206–211.
7. Benayoun, B.A., Dipietromaria, A., Bazin, C. and Veitia, R.A. (2009)
FOXL2: at the crossroads of female sex determination and ovarian
function. Adv. Exp. Med. Biol., 665, 207–226.
8. Uda, M., Ottolenghi, C., Crisponi, L., Garcia, J.E., Deiana, M., Kimber,
W., Forabosco, A., Cao, A., Schlessinger, D. and Pilia, G. (2004) Foxl2
disruption causes mouse ovarian failure by pervasive blockage of follicle
development. Hum. Mol. Genet., 13, 1171–1181.
9. Schmidt, D., Ovitt, C.E., Anlag, K., Fehsenfeld, S., Gredsted, L., Treier,
A.-C. and Treier, M. (2004) The murine winged-helix transcription factor
Foxl2 is required for granulosa cell differentiation and ovary maintenance.
Development, 131, 933–942.
10. Batista, F., Vaiman, D., Dausset, J., Fellous, M. and Veitia, R.A. (2007)
Potential targets of FOXL2, a transcription factor involved in craniofacial
and follicular development, identified by transcriptomics. Proc. Natl
Acad. Sci. USA, 104, 3330–3335.
11. Benayoun, B.A., Batista, F., Auer, J., Dipietromaria, A., L’Ho ˆte, D., De
Baere, E. and Veitia, R.A. (2009) Positive and negative feedback regulates
the transcription factor FOXL2 in response to cell stress: evidence for a
regulatory imbalance induced by disease-causing mutations. Hum. Mol.
Genet., 18, 632–644.
12. Escudero, J.M., Haller, J.L., Clay, C.M. and Escudero, K.W. (2010)
Microarray analysis of Foxl2 mediated gene regulation in the mouse ovary
derived KK1 granulosa cell line: Over-expression of Foxl2 leads to
activation of the gonadotropin releasing hormone receptor gene promoter.
J. Ovarian Res., 3, 4.
13. Pisarska, M.D., Barlow, G. and Kuo, F.-T. (2011) Minireview: roles of the
forkhead transcription factor FOXL2 in granulosa cell biology and
pathology. Endocrinology, 152, 1199–1208.
14. De Baere, E., Beysen, D., Oley, C., Lorenz, B., Cocquet, J., De Sutter, P.,
Devriendt, K., Dixon, M., Fellous, M., Fryns, J.-P. et al. (2003) FOXL2
and BPES: mutational hotspots, phenotypic variability, and revision of the
genotype-phenotype correlation. Am. J. Hum. Genet,, 72, 478–487.
15. De Baere, E., Dixon, M.J., Small, K.W., Jabs, E.W., Leroy, B.P.,
Devriendt, K., Gillerot, Y., Mortier, G., Meire, F., Van Maldergem, L.
et al. (2001) Spectrum of FOXL2 gene mutations in
blepharophimosis-ptosis-epicanthus inversus (BPES) families
demonstrates a genotype–phenotype correlation. Hum. Mol. Genet., 10,
1591–1600.
16. Caburet, S., Demarez, A., Moumne ´, L., Fellous, M., De Baere, E. and
Veitia, R.A. (2004) A recurrent polyalanine expansion in the transcription
factor FOXL2 induces extensive nuclear and cytoplasmic protein
aggregation. J. Med. Genet., 41, 932–936.
17. Nallathambi, J., Moumne ´, L., De Baere, E., Beysen, D., Usha, K.,
Sundaresan, P. and Veitia, R.A. (2007) A novel polyalanine expansion in
FOXL2: the first evidence for a recessive form of the blepharophimosis
syndrome (BPES) associated with ovarian dysfunction. Hum. Genet., 121,
107–112.
18. Beysen, D., Moumne ´, L., Veitia, R., Peters, H., Leroy, B.P., De Paepe, A.
and De Baere, E. (2008) Missense mutations in the forkhead domain of
FOXL2 lead to subcellular mislocalization, protein aggregation and
impaired transactivation. Hum. Mol. Genet., 17, 2030–2038.
19. Dipietromaria, A., Benayoun, B.A., Todeschini, A.-L., Rivals, I., Bazin,
C. and Veitia, R.A. (2009) Towards a functional classification of
pathogenic FOXL2 mutations using transactivation reporter systems.
Hum. Mol. Genet., 18, 3324–3333.
20. Shah, S.P., Ko ¨bel, M., Senz, J., Morin, R.D., Clarke, B.A., Wiegand, K.C.,
Leung, G., Zayed, A., Mehl, E., Kalloger, S.E. et al. (2009) Mutation of
FOXL2 in granulosa-cell tumors of the ovary. N. Engl. J. Med., 360,
2719–2729.
21. Schrader, K.A., Gorbatcheva, B., Senz, J., Heravi-Moussavi, A., Melnyk,
N., Salamanca, C., Maines-Bandiera, S., Cooke, S.L., Leung, P., Brenton,
J.D. et al. (2009) The specificity of the FOXL2 c.402C . G somatic
mutation: a survey of solid tumors. PLoS ONE, 4, e7988.
22. Benayoun, B.A., Caburet, S., Dipietromaria, A., Georges, A., D’Haene,
B., Pandaranayaka, P.J.E., L’Ho ˆte, D., Todeschini, A.-L., Krishnaswamy,
S., Fellous, M. et al. (2010) Functional exploration of the adult ovarian
granulosa cell tumor-associated somatic FOXL2 mutation p.Cys134Trp
(c.402C . G). PLoS ONE, 5, e8789.
23. Kim, J.-H., Yoon, S., Park, M., Park, H.-O., Ko, J.-J., Lee, K. and Bae, J.
(2010) Differential apoptotic activities of wild-type FOXL2 and the
adult-type granulosa cell tumor-associated mutant FOXL2 (C134W).
Oncogene, 30, 1653–1663.
24. Pace, C.N. and Scholtz, J.M. (1998) A helix propensity scale based on
experimental studies of peptides and proteins. Biophys. J., 75, 422–427.
25. Schellman, CG. (1980) The Alpha-L Conformation at the Ends of Helices.
Elsevier/North-Holland Biomedical Press, Jaenicke R, Amsterdam.
26. Saleem, R.A., Banerjee-Basu, S., Murphy, T.C., Baxevanis, A. and
Walter, M.A. (2004) Essential structural and functional determinants
3384Human Molecular Genetics, 2011, Vol. 20, No. 17
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from

Page 10

within the forkhead domain of FOXC1. Nucleic Acids Res., 32,
4182–4193.
27. Saleem, R.A., Banerjee-Basu, S., Berry, F.B., Baxevanis, A.D. and
Walter, M.A. (2003) Structural and functional analyses of disease-causing
missense mutations in the forkhead domain of FOXC1. Hum. Mol. Genet.,
12, 2993–3005.
28. Berry, F.B., Tamimi, Y., Carle, M.V., Lehmann, O.J. and Walter, M.A.
(2005) The establishment of a predictive mutational model of the forkhead
domain through the analyses of FOXC2 missense mutations identified in
patients with hereditary lymphedema with distichiasis. Hum. Mol. Genet.,
14, 2619–2627.
29. Vernes, S.C., Nicod, J., Elahi, F.M., Coventry, J.A., Kenny, N., Coupe,
A.-M., Bird, L.E., Davies, K.E. and Fisher, S.E. (2006) Functional genetic
analysis of mutations implicated in a human speech and language
disorder. Hum. Mol. Genet., 15, 3154–3167.
30. Esteve, J.G. and Falceto, F. (2005) Classification of amino acids induced
by their associated matrices. Biophys. Chem., 115, 177–180.
31. Tanford, C. (1962) Contribution of hydrophobic interactions to the
stability of the globular conformation of proteins. J. Am. Chem. Soc.,
84, 4240–4247.
32. Demarest, S.J., Chung, J., Dyson, H.J. and Wright, P.E. (2002)
Assignment of a 15 kDa protein complex formed between the p160
coactivator ACTR and CREB binding protein. J. Biomol. NMR, 22,
377–378.
33. Obsil, T. and Obsilova, V. (2010) Structural basis for DNA recognition by
FOXO proteins. Biochim. Biophys. Acta, 10.1016/j.bbamcr.2010.11.025.
34. van Rooij, I.A.J., Broekmans, F.J.M., Scheffer, G.J., Looman, C.W.N.,
Habbema, J.D.F., de Jong, F.H., Fauser, B.J.C.M., Themmen, A.P.N. and
te Velde, E.R. (2005) Serum antimullerian hormone levels best reflect the
reproductive decline with age in normal women with proven fertility: a
longitudinal study. Fertil. Steril., 83, 979–987.
35. Higuchi, R., Krummel, B. and Saiki, R.K. (1988) A general method of in
vitro preparation and specific mutagenesis of DNA fragments: study of
protein and DNA interactions. Nucleic Acids Res., 16, 7351–7367.
36. Nishi, Y., Yanase, T., Mu, Y., Oba, K., Ichino, I., Saito, M., Nomura, M.,
Mukasa, C., Okabe, T., Goto, K. et al. (2001) Establishment and
characterization of a steroidogenic human granulosa-like tumor cell line,
KGN, that expresses functional follicle-stimulating hormone receptor.
Endocrinology, 142, 437–445.
37. Joseph, S. and David, R. (2001) Molecular Cloning: A Laboratory
Manual (Third Edition) Cold Spring Harbor Laboratory Press. Cold
Spring Harbor, New York.
38. Hastie, T., Tibshirani, R., Botstein, D. and Brown, P. (2001) Supervised
harvesting of expression trees. Genome Biol., 2, RESEARCH0003.
39. Ellsworth, B.S., Burns, A.T., Escudero, K.W., Duval, D.L., Nelson, S.E.
and Clay, C.M. (2003) The gonadotropin releasing hormone (GnRH)
receptor activating sequence (GRAS) is a composite regulatory element
that interacts with multiple classes of transcription factors including
Smads, AP-1 and a forkhead DNA binding protein. Mol. Cell.
Endocrinol., 206, 93–111.
40. Pannetier, M., Renault, L., Jolivet, G., Cotinot, C. and Pailhoux, E. (2005)
Ovarian-specific expression of a new gene regulated by the goat PIS
region and transcribed by a FOXL2 bidirectional promoter. Genomics, 85,
715–726.
41. Benayoun, B.A., Caburet, S., Dipietromaria, A., Bailly-Bechet, M.,
Batista, F., Fellous, M., Vaiman, D. and Veitia, R.A. (2008) The
identification and characterization of a FOXL2 response element provides
insights into the pathogenesis of mutant alleles. Hum. Mol. Genet., 17,
3118–3127.
42. Kim, H.P., Roe, J.H., Chock, P.B. and Yim, M.B. (1999) Transcriptional
activation of the human manganese superoxide dismutase gene mediated
by tetradecanoylphorbol acetate. J. Biol. Chem., 274, 37455–37460.
Human Molecular Genetics, 2011, Vol. 20, No. 173385
at INSERM on September 7, 2011
hmg.oxfordjournals.org
Downloaded from