ABSTRACT
The thyroid transcription factor-1 homeodomain (TTF-1HD) shows a peculiar DNA binding specificity, preferentially
recognizing sequences containing the 5
'
-CAAG-3
'
core motif. Most other homeodomains instead recognize sites containing the 5
'
-TAAT-3
'
core motif. Here, we show that TTF-1HD efficiently recognizes another sequence, called D1, devoid of the 5
'
-CAAG-3
'
core motif. Different experimental approaches indicate that TTF-1HD contacts the D1 sequence in a manner which is different to that used
to interact with sequences containing the 5
'
-CAAG-3
'
core motif. The binding activities that mutants of TTF-1HD display with the D1 sequence or with the sequence containing the 5
'
-CAAG-3
'
core motif indicate that the role of several DNA-contacting amino acids is different. In particular, during recognition of
the D1 sequence, backbone-interacting amino acids not relevant in binding to sequences containing
the 5
'
-CAAG-3'
core motif play an important role. In the TTF-1HD, therefore, the contribution of several amino acids to DNA recognition
depends on the bound sequence. These data indicate that although a common
bonding network exists in all of the HD/DNA complexes, peculiarities important
for DNA recognition may occur in single cases.
The homeodomain (HD) represents the DNA binding domain of a large number of
transcription factors controlling cell fate decisions (
1
,
2
). Different HDs show a similar structure consisting of three helical regions
(I, II and III) folded into a tight globular structure (
3
). Helix I is preceded by an N-terminal arm and separated by a loop from helix II, which forms, with
helix III, a helix-turn-helix motif (H-T-H). The latter has previously been described for
several prokaryotic gene-regulatory proteins (
4
). However, differently from prokaryotic H-T-H motifs, in which dimerization is required for high affinity DNA
binding, most HD bind with high affinity to DNA as monomers (
5
). Only for the
paired
(prd) class of HDs has a cooperative dimerization been observed (
6
-
7
). The DNA binding mode of HDs is highly conserved (
3
,
8
-
9
). Helix III (also called the recognition helix) lies in the major groove of the
DNA, where it establishes specific contacts to bases. Additional specific
contacts to bases are made by the N-terminal arm (in the minor groove). The loop between helices I and II
interacts with the DNA backbone.
The amino acids which control the DNA binding specificity of HDs are located
mostly in the recognition helix and in the N-terminal arm (
10
-
21
). However, the relative importance of each residue appears to be different,
depending on the HD. For example, only a functional comparison of fushi tarazu
(ftz) and muscle cell homeobox (msh) HDs has revealed that the amino acids at
positions 28 and 43 could be relevant in the control of DNA binding specificity
(
18
). These observations could suggest that HD-DNA interactions, although based on a common bonding network, could
differ in some details sometimes important for DNA recognition. These
differences may occur even when the same protein interacts with different DNA
sequences. In fact, it has been demonstrated that the binding specificity of
the prdHD is controlled by amino acids at the N-terminus of the recognition helix only in interaction with particular
sequences (
14
,
22
).
The HD of thyroid transcription factor-1 (TTF-1HD) has a structure similar to other HDs (
23
) but shows a peculiar DNA binding specificity. In fact, TTF-1HD preferentially recognizes sequences having the 5'-CAAG-3' core motif and, in contrast to other HDs, it
binds only with low affinity sequences containing the 5'-TAAT-3' core motif (
17
). In this study we show that TTF-1HD is able to specifically recognize a sequence devoid of the 5'-CAAG-3' core motif. By means of different experimental
approaches we demonstrate that TTF-1HD interacts with this sequence by using a binding mode which is
different to that used to recognize sequences containing the 5'-CAAG-3' core. Backbone-interacting amino acids not relevant in binding
to sequences containing the 5'-CAAG-3' core motif play an important role in recognizing the
sequence devoid of the 5'-CAAG-3' core motif.
Construction of the bacterial expression vectors pT7.7 TTF-1HD and pT7.7 TTF-1HD(K
50
), pT7.7 TTF-1HD(6N[Delta]) and M2HD has been described elsewhere (
17
,
24
). Mutants TTF-1HD(R
28
) and M2HD(R
28
) have been constructed by the method of Ho
et al
. (
25
) and cloned in the vector pT7.7. Proteins were expressed using the BL21 (DE3)
Escherichia coli
strain (
26
). The expressed proteins were purified essentially as described (
5
) with two chromatographic steps using Biorex and Mono S ion exchange matrices.
The purity of the proteins was checked by SDS-PAGE and was >= 99%. The sequences of the oligonucleotides used in gel retardation
and methylation interference assays were as follows (top strand): C, 5'-CACTGCCCAGTCAAGTGTTCTTGA-3'; C
ant1
, 5'-CACTGCCCAGTTAATTGTTCTTGA-3'; D1, 5'-ACGATGAGTGGCTCATAAATCG-3'; D1 33mer, 5'-CGATCACGATGAGTGGCTCATAAATCGCATGCT-3'.
Oligonucleotides were labeled at the 5'-end using polynucleotide kinase and [[gamma]-
32
P]ATP and annealed with the complementary strand.
Gel retardation assays were performed by incubating protein and DNA in a buffer
containing 20 mM Tris-HCl, pH 7.6, 75 mM KCl, 0.25 mg/ml bovine serum albumin (BSA), 5 mM
dithiothreitol (DTT), 5 [mu]g/ml calf thymus DNA, 10% glycerol for 30 min at room temperature. Calf
thymus DNA was omitted in gel retardation assays to measure equilibrium
dissociation constants (
K
d
). Protein-bound and free DNA were separated on native 7% polyacrilamide gels run in
0.5* TBE for 2 h at 4oC. Gels were dried, exposed to X-ray films and the bands quantitated by densitometric scanning
of the autoradiogram using a LKB laser densitometer.
K
d
values were calculated as in Damante
et al
. (
27
). The DNA binding activities of HDs shown in Figure
5
were determined by calculating the ratio of protein-bound and free DNA signals using the same amount of protein. The values
were then normalized to the value obtained for the complex TTF-1HD/C sequence, arbitrarily considered as 100.
Methylation interference experiments were done according to Zannini
et al
. (
28
), using as probes dimethylsulfate-treated D133mer oligonucleotides. Protein-bound and free DNAs were separated by preparative PAGE and recovered
as described (
29
). After the chemical cleavage the products were separated on 20% denaturing
polyacrylamide gels and visualized by autoradiography.
Spectra were recorded in a 2 mm cell at 10 [mu]M DNA duplex and 10 [mu]M protein concentrations in 5 mM NaH
2
PO
4
, 5 mM MgCl
2
, 50 mM NaClO
4
, pH 7.5, at 25oC. Data were collected on a Jasco J-600 spectropolarimeter (Jasco Inc., Easton, MD). The averages of 10
scans were baseline-corrected with a spectrum of buffer alone and were smoothed using software
provided by Jasco. Difference spectra were calculated as described (
30
,
31
). The [alpha]-helical content was calculated according to Menendèz-Arias
et al
. (
32
).
The D1 sequence has been identified by site selection experiments starting from
rat genomic DNA (
33
; Fabbro
et al
., unpublished data). A measure of the equilibrium dissociation constants (
K
d
) of the TTF-1HD/D1, TTF-1HD/C and TTF-1HD/C
Ant1
complexes was performed (Fig.
1
). The C sequence is a high affinity TTF-1HD binding site containing the 5'-CAAG-3' core motif. The C
ant1
sequence is derived from the C sequence by changing the 5'-CAAG-3' motif to 5'-TAAT-3'. As such, C
Ant1
is a low affinity binding site for TTF-1HD (
22
). The apparent
K
d
for the TTF-1HD/D1 complex is similar to that measured for the TTF-1HD/C complex (0.34 * 10
-9
and 0.22 * 10
-9
M respectively) and much lower than that observed for the TTF-1HD/C
Ant1
complex (0.38 * 10
-8
M). These data demonstrate that the D1 sequence, though not containing a 5'-CAAG-3' core motif (see D1 sequence in Fig.
3
B), is a high affinity binding site for TTF-1HD.
The lack of significant homology between the C and D1 oligonucleotides suggests
that TTF-1HD could recognize these sequences by using different binding modes. This
possibility is also suggested by the different migration rates observed for the
TTF-1HD/C and the TTF-1HD/D1 complexes in gel retardation assays. Figure
2
shows that although the free 22mer D1 oligonucleotide migrates faster than the
free 24mer C oligonucleotide, the TTF-1/D1 complex migrates slower than the TTF-1/C complex. These results are compatible with different structural
conformations of the protein/DNA complexes resulting in different hindrances
during gel migration. It is important to note that the complexes established by
TTF-1HD with other oligonucleotides containing the 5'-CAAG-3' motif but having either a different length
(18mer or 14mer) or base mutations show a migration rate equal to that observed
for the TTF-1HD/C complex (data not shown). We can exclude the possibility that the
slow migration rate of the TTF-1HD/D1 complex is due to dimerization because when two TTF-1HD molecules bind to the same DNA molecule the ternary complex show
a much slower migration rate (data not shown).
The data described above indicate that TTF-1HD interacts with the D1 sequence in a manner which is different to that
used to interact with the C sequence. Based on these data we wanted to address
the question of whether critical amino acids for specificity in the TTF-1HD/C interaction would play a similar role in the TTF-1/D1 interaction. Figure
5
shows the binding activity of several TTF-1HD mutants with the C, C
Ant1
and D1 sequences.
Mutant TTF-1HD(6N[Delta]) lacks the N-terminal arm, which is known to contribute to the interaction
when the protein binds the C site (
17
). The binding activity of TTF-1HD(6N[Delta]) is low with all sequences, indicating that the N-terminal arm is essential to establish an efficient
interaction with either the C sequence or with the D1 sequence.
In mutant M2HD the TTF-1HD N-terminus of the recognition helix has been mutagenized. Pro
42
, Thr
43
and Val
45
have been changed to Glu, Arg and Ile respectively (
24
). In the C context, mutant M2HD recognizes the 5'-CAAG-3' but not the 5'-TAAT-3' motif and, therefore,
behaves as the wild-type TTF-1HD. However, M2HD shows very much reduced binding activity with the
D1 sequence, indicating that the N-terminal region of the recognition helix (containing Pro
42
, Thr
43
and Val
45
) plays a significant role in recognition of this sequence.
One of the amino acids controlling TTF-1HD binding specificity is Gln
50
(
17
). When Gln
50
is changed to Lys [mutant TTF-1HD(K
50
)], binding activity to the sequence 5'-CAAG
In the context of the C sequence, the amino acid located at position 54 plays a
role in binding specificity. In fact, when the wild-type Tyr
54
is changed to Met, binding activity to the C sequence is reduced while, in
contrast, binding activity to the C
Ant1
sequence is increased (
21
). TTF-1HD(M54) binds D1 as well as wild-type TTF-1HD, indicating that, in the context of the D1 sequence, the
side chain of the amino acid at position 54 is less important than in the
context of the C sequence.
Amino acids at positions 50 and 54 of HDs contact bases in the major groove (
3
). The binding activities of both TTF-1HD(K
50
) and TTF-1HD(M
54
) indicate that in the TTF-1HD/D1 interaction the contacts to bases established by amino acids at
position 50 and 54 play a role much less important with respect to that played
in the TTF-1HD/C interaction. In contrast, the binding activity of M2HD indicates
that in the context of D1 sequence amino acids at the N-terminus of the recognition helix appear to be important. In Mat[alpha]2/DNA and Antp/DNA complexes amino acids at positions 42 and 43
respectively establish contacts with the sugar-phosphate backbone (
9
,
37
). Therefore, the different binding activities shown by TTF-1HD and M2HD with the C and D1 sequences would indicate a significant role
of backbone contact(s) in the TTF-1HD/D1 interaction but not in the TTF-1HD/C interaction.
The amino acid at position 43 plays a role in the differential DNA binding
properties observed for ftzHD and mshHD, in combination with the amino acid at
position 28 (
18
). Interestingly, TTF-1HD possesses Ala and Thr respectively at positions 28 and 43. This
configuration is very similar to that of mshHD (Ile
28
and Thr
43
). However, M2HD possesses an Arg at position 43, as does the ftzHD. These
observations induced us to test the relevance of the amino acid at position 28
of TTF-1HD. In place of the naturally occurring Ala
28
, we introduced Arg, which instead is present at position 28 of ftzHD. Arg
28
was introduced into both TTF-1HD and M2HD, giving rise to TTF-1HD(R
28
) and M2HD(R
28
) respectively. Figure
5
shows that the binding activity of TTF-1HD(R
28
) is only moderately reduced compared with that of TTF-1HD with the C sequence. In contrast, TTF-1HD(R
28
) interacts much less efficiently than TTF-1HD with the D1 sequence. These results indicate that the amino acid at
position 28 plays an important role in DNA recognition only in the context of
the D1 sequence, and not in that of the C sequence. The binding activity of
mutant M2HD(R
28
) with the C sequence is reduced to 40%, compared with TTF-1HD, but to 5% with the D1 sequence. In the Antp/DNA complex Arg
28
establishes an electrostatic interaction with the phosphate backbone (
37
). Therefore, our results indicate that backbone interactions (established by
amino acids at positions 28 and 43) play an important role in recognition of
the D1 sequence, but only a marginal role in recognition of the C sequence. In
contrast, results obtained with mutants TTF-1HD(K
50
) and TTF-1HD(M
54
) indicate that base contacts established by amino acids at positions 50 and 54
play a role in C recognition, but not in D1 recognition. These data are
summarized in Figure
6
.
Our data demonstrate that TTF-1HD is able to recognize with high affinity a sequence devoid of the 5'-CAAG-3' core motif. In such a way, the DNA binding
specificity of this protein appears to be wider than previously appreciated.
Results obtained for other HD-containing proteins, using either
in vivo
or
in vitro
binding assays, indicate that other HDs are also able to recognize a broad
spectrum of DNA sequences (
38
-
39
). Therefore, the ability to recognize different DNA sequences appears to be a
common feature of HDs.
The binding mode of TTF-1HD and, in turn, the importance of contact amino acids may change in
recognizing different DNA sequences. The biological implications of this
phenomenon are evident: the nature of a contact amino acid would be relevant
for the recognition of only a subset of controlled genes. Such a possibility is
compatible with results obtained
in vivo
with the ftz protein (
40
). In fact, by using
Drosophila
mutants defective in the ftz gene, it has been clearly demonstrated that when a
ftz mutation in which the Gln
50
of ftz is changed to Lys is introduced, the mutant protein is able to provide
an efficient rescue of parasegments 8 and 14, but a much poorer rescue of other
structures. In view of our findings, the Lys
50
-containing ftz mutant would rescue only the activity of genes controlled
by DNA sequences for which the nature of the amino acid at position 50 of ftz
protein is not relevant for an efficient interaction. The versatility of HDs in
recognizing DNA sequences is one of the most important features to explain the
success of these proteins during evolution (
41
).
Although several studies have revealed that the binding mode of HDs is conserved
overall, the present data demonstrate that in each HD/DNA complex peculiarities
may occur and play a role in DNA recognition. To fully understand the molecular
bases of the biological activity of HD-containing proteins, it might be of relevance to reveal these
peculiarities.
The first two authors (D.F. and G.T.) have contributed equally to this work.
This research is funded by a grant to G.D. from the Associazione Italiana per
la Ricerca sul Cancro (AIRC). D.F. is also supported by a AIRC fellowship. We
gratefully acknowledge R. Di Lauro for helpful discussions, L. Minichiello for
suggestions in writing the manuscript and L. Cattarossi for computational
support.
REFERENCES
Return