1
SUPPLEMENTAL INFORMATION
Steric mechanism of auto-inhibitory regulation of specific and non-specific
DNA binding by the ETS transcriptional repressor ETV6
Soumya De1, Anson C. K. Chan2, H. Jerome Coyne III1, Niraja Bhachech3,4, Ulrike Hermsdorf1,
Mark Okon1, Michael E. P. Murphy2, Barbara J. Graves3,4 and Lawrence P. McIntosh1*
1 Department of Biochemistry and Molecular Biology, Department of Chemistry, and Michael
Smith Laboratories, University of British Columbia, Vancouver BC, V6T 1Z3, Canada
2 Department of Microbiology and Immunology, University of British Columbia, Vancouver BC,
V6T 1Z3, Canada
3 Department of Oncological Sciences, University of Utah School of Medicine, Huntsman Cancer
Institute, University of Utah, Salt Lake City, UT, 84112-5550, USA
4 Howard Hughes Medical Institute, Chevy Chase, MD, 20815-6789, USA
2
3
Supplemental Fig. S1. HSQC spectra of ETV6.
15N-HSQC spectra of ETV6D446 (red), ETV6D446–DNAsp (green) and ETV6D446–DNAnonsp
(magenta). The latter were recorded with protein in the presence of a 1.1 fold molar excess of
DNAsp and 3 fold molar excess of DNAnonsp, respectively. The cluster of sharp peaks with 1HN
shifts of ~ 8 to 8.5 ppm in the spectra of the two DNA complexes, but not of the free protein,
arise from amides in the unfolded helix H5. Also shown are the spectra of the two complexes
superimposed upon that of ETV6R426 (cyan) to highlight changes associated with DNA binding
and not helix unfolding. Peaks highlighted in Figure 1A are labeled.
4
5
Supplemental Fig. S2. Dynamic properties of the ETV6D446’-DNAsp complex probed by 15N
relaxation measurements.
Shown are the amide 15N R1, R2 and heteronuclear {1H}-15N NOE data recorded for the
perdeuterated ETV6D446’–DNAsp complex at 25 oC with a 600 MHz spectrometer.1 These data
were fit to the model-free formalism2 with the program TENSOR23 to yield a global isotropic
tumbling correlation time of 12.5 ± 0.2 nsec and the plotted conventional order parameters (S2).
For residues 339 to 424, the S2 values were obtained using an anisotropic diffusion tensor and
the ETV6R426–DNAsp-cryst crystal structure, whereas an isotropic model was used for the
remaining C-terminal residues. With relatively slow R2 relaxation rate constants and low 15N-
NOE and S2 values, the residues corresponding to helix H5 are clearly more mobile than those
in the well-ordered ETS domain. Along with their random coil chemical shifts (Fig. 2A), these
data collectively show that this helix is unfolded in the ETV6D446–DNAsp complex.
The secondary structural elements of ETV6 are shown on top (α-helices as ellipses and
β-strands as arrows). The core ETS domain is colored in red, helix H4 of CID is in cyan, and
residues corresponding to helix H5 in the free protein are represented as a white ellipse and
highlighted with a gray shaded rectangle. Missing data correspond to prolines, residues with
overlapping signals, or residues not adequately fit by this formalism.
6
Supplemental Fig. S3. NMR-monitored titration curves of ETV6D446 with DNAnonsp.
15N-labeled ETV6D446 (initial 0.2 mM, final 0.15 mM) was titrated with aliquots of a 1.8
mM stock solution of DNAnonsp in sample buffer to molar ratios 0, 0.1, 0.25, 0.5, 0.75, 1, 2 and 3.
See Figure 1 and Supplemental Figure S1 for the original 15N-HSQC spectra. The changes in
combined amide chemical shifts (Δδ = {(ΔδH2 + (0.154ΔδN)2}1/2) for several residues are plotted
as a function of DNAnonsp to ETV6D446 molar ratio. The flattening of the titration curves indicates
that the protein is essentially saturated at the final titration points with a 2- or 3-fold molar
excess of DNAnonsp. Under these experimental conditions, the NMR data could not be fit reliably
to obtain a KD value due to the small amount of free versus total DNAnonsp at the initial titration
points. Lower concentrations were used for the ITC studies shown in Figure 4.
0
0.04
0.08
0.12
0 1 2 3 4
I366
0
0.05
0.1
0.15
0.2
0 1 2 3 4
S391
T418
0
0.04
0.08
0.12
0 1 2 3 4
Y387
0
0.05
0.1
0.15
0.2
0.25
0 1 2 3 4
[DNAnonsp] / [ETV6D446]
Dd (
ppm
)
7
Supplemental Fig. S4. Effect of helix H5 on the chemical shifts of ETV6 ETS domain.
(A) Overlaid 15N-HSQC spectra of inhibited ETV6D446 (red) and uninhibited ETV6R426
(blue). (B) Changes in combined amide chemical shifts (Δδ = {(ΔδH2 + (0.154ΔδN)2}1/2) of the
corresponding backbone amides between the two fragments mostly occur on the DNA binding
interface. This is due to the deletion of residues 427-446, which include the CID helix H5 that
packs along helix H3 in free ETV6D446. Note that the structural ensembles of ETV6R426 and
ETV6D446 overlap closely (see Figure 3B, main text), indicating that these amide shift changes
reflect subtle perturbations. The secondary structure (helix, cylinder; strand, arrow) of ETV6R426
is displayed in the top cartoon with the core ETS domain in red and helix H4 of the CID in cyan.
8
9
Supplemental Fig. S5. Comparison of the magnitudes and signs of the amide chemical
shift changes accompanying formation of specific versus non-specific complexes.
Here we compare the "direction" and magnitude of the amide chemical shift
perturbations (CSP's) for the specific (ETV6D446–DNAsp) and non-specific (ETV6D446–DNAnonsp)
complexes with respect to the free uninhibited protein (ETV6R426). The secondary structure
elements of the protein are shown on top and the DNA-binding interface identified in the
ETV6R426–DNAsp-cryst crystal structure is highlighted in grey. (A, B) In contrast to the combined
amide CSP's shown in Fig. 1, separate CSPs (Δδ) for the amide 15N and 1HN signals in each
complex versus the free protein are presented. For both complexes, the majority of these
perturbations are localized in the same general binding interface. Furthermore, for most
residues, the changes are also of the same sign (i.e., positive or negative) for both complexes.
(C) The relative magnitude and direction of the CSP's for each residue can also be compared
by a projection analysis.4 For this analysis, the CSP's for ETV6D446–DNAsp and ETV6D446–
DNAnonsp versus ETV6R426 are represented as vector quantities (inset cartoon). The projections
of the vectors for the non-specific complex along the corresponding vectors for the specific
complex are shown by blue bars. The cosines of the angles between the two vectors are
shown by red diamonds. For most residues, cos() ~ 1 but the projection is < 1. This indicates
that their amide 15N and 1HN signals shift in the same "direction," but with less magnitude, upon
binding DNAnonsp relative to DNAsp. However, some amides shift differently. Collectively, these
analyses support the conclusion that ETV6 binds non-specific and specific DNA by a similar
canonical interface of the ETS domain. However, the interactions are not identical and are
generally stronger for DNAsp.
Note that these chemical shifts are for ETV6D446 in the presence of a 1.1-fold molar
excess of DNAsp and 3-fold molar excess of DNAnonsp, and thus correspond to 99% and 98%
saturation, respectively, based on their respective KD values and binding site sizes. Missing data
correspond to prolines, residues without confident assignments, and in the case of the
projection analysis, residues with combined Δδ less than a cut-off of 0.03 ppm.
10
Supplemental Fig. S6. His396 establishes the specificity of ETV6 for 5'GGAA3' sequences.
The KD values of WT- and H396Y-ETV6D446 for a DNA binding site containing a core
5'GGAA3' versus 5'GGAT3' were measured by an electrophoretic mobility shift assay (EMSA).
Both auto-inhibited proteins bind the 5'GGAA3' oligonucleotide with similar affinities. However,
the WT protein displays a 500-fold preference for the 5'GGAT3' over the 5'GGAA3' binding site. In
contrast, the affinity of H396Y-ETV6D446 displays only an 8-fold preference.
The EMSA assays were carried out as described previously,5 using samples in 25 mM
Tris pH 7.9, 60 mM KCl, 0.6 mM MgCl2, 0.1 mM EDTA, 10 mM DTT, 10% glycerol, 0.5 μg/mL
BSA and 2.5 ng/μL poly(dIdC) at 4oC. The data points and associated error bars were obtained
from three replica experiments, and fit to a simple binding isotherm with KaleidaGraph, as
previously described. The oligonucleotide sequences (top strand) of the two DNA binding sites
were:
5'TCGACGGCCAAGCCGGAAGTGAGTGCC3'
5'TCGACGGCCAAGCCGGATGTGAGTGCC3'
0
0.2
0.4
0.6
0.8
0
0.2
0.4
0.6
0.8
10-12 10-11 10-10 10-9 10-8 10-7 10-6 10-5 10-12 10-11 10-10 10-9 10-8 10-7 10-6 10-5
fra
ction
bo
und
[protein] (M)
WT-ETV6D446 H396Y-ETV6D446
GGAT
GGAT
1.0 1.0
GGAAKD = 1.6 ± 0.3 nM
GGAAKD = 1.9 ± 0.3 nM
KD = 800 ± 100 nM
KD = 16 ± 7 nM
11
Supplemental Fig. S7. Thymine at +4 is unlikely to hydrogen bond with His396.
Unlike most ETS factors that bind either 5'GGAT3' or 5'GGAA3', ETV6 preferentially binds
the latter due to His396 interacting with the complementary base T+4’. (A) Tautomeric states of
the neutral histidine side chain. In the free amino acid, the Nε2H tautomer is favored by ~5-fold.6
(B) Upper: In the crystal structure of the ETV6R426–DNAsp-cryst complex, the His396 Nε2 is within
hydrogen bonding distance (2.8 Å) of the O4 of T+4’ on the complimentary DNA strand. Note that
the side chain of His396 must be in the positively-charged imidazolium or neutral Nε2H imidazole
tautomeric state to act as a hydrogen bond donor. Lower: The nucleotide bases at position
+4/+4' of the ETV6D446-DNAsp complex were modeled as the T-A pair. This would result in the
loss of the hydrogen bond between His396 and T+4’. Furthermore, no new hydrogen bonds
could form, unless perhaps the imidazole is in the neutral Nδ1H tautomeric state, thereby
allowing Nε2 to act as a hydrogen bond acceptor with the exocyclic amine of A+4’. (C) In contrast,
12
many ETS factors have a tyrosine residue at the position corresponding to His396, and bind
either 5'GGAT3' or 5'GGAA3'. For example, in the crystal structures of ELK4 (or SAP-1) bound to
the E74 (upper: TACCGGAAGT; 1BC8.pdb) and the c-fos (lower: GACAGGATGTG; 1BC7.pdb)
promoter sequences, Tyr65 (crystal structure numbering) interacts with the exocyclic amine of
an adenine on either DNA strand.7 Water molecule is shown as a blue circle. In B and C:
carbon, green; oxygen, red; nitrogen, blue.
Supplemental References
1. Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal, S. M., Kay, C. M., Gish, G., Shoelson, S. E., Pawson, T., Forman-Kay, J. D. & Kay, L. E. (1994). Backbone dynamics of a free and phosphopeptide-complexed Src homology 2 domain studied by 15N NMR relaxation. Biochemistry 33, 5984-6003.
2. Lipari, G. & Szabo, A. (1982). Model-free approach to the interpretation of nuclear magnetic
resonance relaxation in macromolecules. 2 - Analysis of experimental results. J. Am. Chem. Soc. 104, 4559-4570.
3. Dosset, P., Hus, J. C., Blackledge, M. & Marion, D. (2000). Efficient analysis of
macromolecular rotational diffusion from heteronuclear relaxation data. J. Biomol. NMR 16, 23-28.
4. Selvaratnam, R., Mazhab-Jafari, M. T., Das, R. & Melacini, G. (2012). The auto-inhibitory
role of the EPAC hinge helix as mapped by NMR. Plos One 7, e48707. 5. Green, S. M., Coyne, H. J., 3rd, McIntosh, L. P. & Graves, B. J. (2010). DNA binding by the
ETS protein TEL (ETV6) is regulated by autoinhibition and self-association. J. Biol. Chem. 285, 18496-504.
6. Blomberg, F., Maurer, W. & Ruterjans, H. (1977). Nuclear magnetic resonance
investigation of 15N-labeled histidine in aqueous solution. J. Am. Chem. Soc. 99, 8149-59. 7. Mo, Y., Vaessen, B., Johnston, K. & Marmorstein, R. (1998). Structures of SAP-1 bound to
DNA targets from the E74 and c-fos promoters: insights into DNA sequence discrimination by Ets proteins. Mol. Cell 2, 201-12.