Supplemental information.
Expression and purification of HIFs and ARNT PasB domain proteins.
Expression and purification of HIF-1 R245E (947 construct) and R245E/E266H/R311H/S330L
(1106 construct) mutants and HIF-2 R247E (949 construct) for ITC, SPR and NMR studies was
performed as describe previously [24]. Additionally, 6xHis-GST-HIF-1 E245R construct was
generated for purification and crystallization of HIF-1 R245E/ARNT E362R PasB heterodimer.
Tobacco Etch Virus (TEV) protease site was inserted after GST in this construct. To generate HIF1 and ARNT proteins suitable for AlphaScreen assay development, the Avi tag sequence
(LNDIFEAQKIEWHE) and FLAG tag (DYKDDDDK) were fused to the C-terminus of HIF-1 and
ARNT PasB domains after a Gly-Gly-Gly linker. For expression of HIF-1 PasB biotinylated
protein, the AviTag-containing plasmid was co-transformed with a plasmid containing biotin
ligase. 10 M biotin was added to the growth media 30 min prior induction of protein
expression. Biotinylated protein was purified as described and its identity was confirmed by LCMS analysis. All constructs used in this study are summarized in Table S-1.
Alpha-ELISA Assay development.
The AlphaScreen technology is based on transfer of singlet oxygen from donor to acceptor bead
types when brought into proximity from binding of compatible components immobilized onto
respective bead types. We generated biotinyl-HIF-1 R245E and Flag-ARNT E362R reagents for
use with AlphaScreen streptavidin-coated donor beads and anti-Flag acceptor beads,
respectively. The association of biotinyl-HIF-1 R245E with Flag-ARNT E362R mediated by the
AlphaScreen beads is concentration and time dependent and yields an apparent Kd for binding
of ca. 5nM for each protein (data not shown).
The AlphaScreen assay technology was implemented as a high throughput methodology to
screen for modulators of HIF-1/ARNT PPI. Small molecules that directly prevent protein
association, or HIF proteins that sequester the HIF-1 or ARNT binding partner impart a
reduction in the AlphaScreen signal. The later application allowed us rapidly screen various
protein constructs and protein-compound conjugates used in this study using micro scale
protein expression and purification protocols. In the absence of tool compounds, HIF PasB
domains were used in competition studies to validate the utility of the assay to identify PPI
inhibitors.
To demonstrate that the assay signal is derived from viable PPI interaction and not non-specific
bead association, competition studies were done using proteins that lack capture tags for
AlphaScreen beads. Protein HIF-1 R245E inhibited the AlphaScreen signal generated from
5nM concentrations of biotinyl-HIF-1 R245E and Flag-ARNT E362R with an IC50 value of 43 ±
15nM (n=6). Similarly, soluble ARNT E362R yields and IC50 of 12 ± 8nM (n=12). HIF-1 1106
protein inhibited association of biotinyl-HIF-1 and Flag-ARNT in the AlphaScreen assay.
Identification of reactive compounds using Matrix-assisted laser desorption/ionization timeof-flight mass spectrometry (MALDI-TOF MS).
The matrix employed was 2,5-dihydroxyacetophenone (DHAP) prepared by dissolving 5mg
DHAP in 250L ethanol and adding 80L of diammonium citrate dissolved in ultrapure water.
This matrix produced protein MS data without matrix adducts that can give rise to interfering
peaks, thus yielding easily interpretable protein-compound MS spectra (data not shown). Two
L of protein-compound sample solution was mixed with 2L of a 2% trifluoroacetic acid
solution (aq) and 0.5L of the mixture was spotted on a polished stainless steel MALDI plate.
Mass spectra were acquired using an automated acquisition and spectral processing method
(AutoExecute™) in linear mode with an accelerating voltage of 20kV. The instrument was
calibrated with masses derived from standard protein (Bruker Protein Standard II) spectra. 500
mass spectra were summed from each sample.
Nanoelectrospray liquid chromatography tandem mass spectrometry (nanoLC-MS/MS).
Compounds that demonstrated >50 % reactivity (defined as intensity of reacted protein in MS
spectrum/intensity of unreacted protein peak) at the 24h time point were subjected to limited
proteolysis and nanoLC-MS/MS in a linear ion trap (Thermo LTQ) to localize the fragment
binding site (Table 2). Volumes of reaction mixtures corresponding to 10g protein were
denatured by heat and the addition of 10% (volume) acetonitrile and digested with trypsin
(1:10 enzyme:protein) overnight at 37°C. Dried digests were reconstituted in 15L 0.1% formic
acid (aq) and 3L samples were analyzed by MS/MS analysis. MS/MS data were analyzed with
Spectrum Mill™ software to extract the molecular features of each sample. Experimental LCMS/MS data sets were searched against the theoretical MS/MS spectra derived from predicted
tryptic peptides from the protein construct sequence, with the addition of the test compound
masses on Cys, His and Lys residues as possible chemical modifications.
Table S-1. HIF-1 and ARNT constructs.
Target
Construct
Domain
number
boundary
HIF-1
900
238-349
WT
AlphaScreen
polyHis
---
HIF-1
947
238-349
R245E
SPR
polyHis
---
HIF-1
1383
238-349
R245E
Protein
polyHis-
crystallization
GST
R245E, E266H,
MS reactive
polyHis
---
R311H, S330L
screen
HIF-1
1106
238-349
Mutations
Assay
N-term
C-term Tag
Tag
HIF-1
1215
238-349
R245E
AlphaScreen
polyHis
Biotin
ARNT
829
356-470
WT
AlphaScreen
polyHis
---
ARNT
948
356-470
E362R
SPR, protein
polyHis
---
polyHis
FLAG
crystallization
ARNT
1102
356-470
E362R
AlphaScreen
Table S-2. AlphaSreen analysis of HIF-2 to HIF-1 mutants.
PAS-B
Protein
Construct
Domain
Boundary
Mutation
AlphaScreen
IC50 nM ± sem
HIF-1
947
238-349
R245E
35 ± 2
HIF-1
1518
238-349
R245E , P329T
2580 ± 180
HIF-1
1519
238-349
R245E , M338V
220 ± 10
HIF-1
1520
238-349
R245E, L344V
1760 ± 200
HIF-2
949
240-350
R247E
890 ± 350
Fig.S-1. Superposition of Hif-1/ARNT (orange/green ribbons) and Hif-2/ARNT (magenta/blue
ribbons).
Time (min)
-10
0.05
0
10
20
30
40
50
60
µcal/sec
0.00
µcal/sec
Time (min)
-10 0 10 20 30 40 50 60 70 80 90
0.05
70
-0.05
0.00
-0.05
2.5
kcal/mole of injectant
kcal/mole of injectant
-0.10
3.0
2.0
1.5
1.0
0.5
0.0
-0.5
0.0
0.5
1.0
Molar Ratio
1.5
2.0
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.0
0.5
1.0
1.5
Molar Ratio
2.0
Fig. S-2. ITC results for HIF-1-ARNT complex at 10 oC . Left: Isotherm of ARNT binding HIF-1
(947) with an observed KD of 125 nM, stoichiometry of 0.97, and H of 2.4 kCal mol-1 G -8.9
kCal mol-1 and TS of 11.3 kCal mol-1. Right: Isotherm of ARNT binding stabilized HIF-1 (1106)
with an observed KD of 83nM, n = 0.97, and H of 1.1 kCal mol-1 G -9.2kCal mol-1 and TS of
10.3 kCal mol-1. The modest increase in affinity of the stabilized HIF-1-ARNT complex is the
result of a lower H penalty of binding.
-10
0
10
Time (min)
20 30 40
50
60
0
70
Time (min)
10 20 30 40 50 60 70 80 90
0.20
0.05
0.15
µcal/sec
µcal/sec
0.10
0.00
-0.05
0.05
0.00
-0.05
-0.10
3.50
-0.10
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
KCal/Mole of Injectant
kcal/mole of injectant
3.00
0.0
0.5
Molar Ratio
1.0
1.5
2.50
2.00
1.50
1.00
0.50
0.00
-0.50
0.0
0.5
1.0
1.5
2.0
2.5
Molar Ratio
Fig. S-3. ITC results for HIF-2-ARNT complex at 10 oC . Left: Isotherm for ARNT binding HIF-2
(949) with an observed KD of 1400 ± 100 nM, n = 0.95, and H of 3.8 ± 1.1 kCal mol-1, G -7.6
kCal mol-1 and TS of 11.4 kCal mol-1. Right: Isotherm for ARNT binding HIF-2 M338V (1519)
with an observed KD of 71 ± 35nM, stoichiometry of 0.99, and H of 2.8 ± 0.1kCal mol-1, G -9.3
kCal mol-1 and TS of 12.1 kCal mol-1. The significant increase in affinity observed for the M338V
HIF-2-ARNT complex, results from a combination of less unfavorable H and more favorable
S of binding. Data for the HIF-2M338V-ARNT complex at 20 and 30 oC (not shown) reveal a
heat capacity change (CP) of ~ -0.18 kCal mol-1 K-1. This is consistent with the burial of
significant non-polar surface area, observed in the crystal structure of the complex.
Figure S-4. MS-MS spectrum for the HIF-1 1106 tryptic peptide (m/z=565.6) incorporating
conjugation of COMPOUND 5 on Cys255. Full sequence coverage confirmed Cys255
conjugation, while the unmodified peptide form of Cys255 was not observed.