Structure of an HIF-1–pVHL Complex: Hydroxyproline Recognition in Signaling Jung-Hyun Min, Haifeng Yang, Mircea Ivan, Frank Gertler, William G. Kaelin Jr., Nikola P. Pavletich

Crystallization and structure determination. The VBC complex was prepared as previously described (S1). The HIF-1 peptide(VBC complex was prepared by mixing purified VBC with a one molar ratio of HIF-1 peptide determined by a native-gel-based titration, followed by gel-filtration chromatography. The final protein concentration used in the crystallization experiments was ~21 mg/ml in 5 mM bis-tris propane (pH 7.0), 200 mM NaCl, and 5 mM DTT. Crystals were grown at 25°C by the hanging-drop vapor-diffusion method, from 24% (w/v) PME 5000, 0.1 M potassium phosphate (pH 6.5), 0.2 M ammonium sulfate and 5 mM DTT. They form in space group P4₃2₁2, with a = b = 59.4 Å and c = 243.9 Å, and contain one HIF-1(VBC quaternary complex in the asymmetric unit. Data were collected at the F1 beamline of Cornell High Energy Synchrotron Source (MacCHESS) with crystals flash-cooled to -160°C in stabilization buffer supplemented with 15% glycerol. Data were processed using the HKL suite (S2). The structure of the VBC complex was determined by the molecular replacement method using the apoVBC structure (S1) as a search model with AMORE (S3). The HIF-1 model was built into the model-phased maps with the program O (S4) and refined by CNS (S5) at 1.85 Å resolution (Table 1). The electron density of HIF-1 is shown in fig. S1 and the sequence alignment of the pVHL domain in fig. S2. The following residues have poor electron density and are presumed to be disordered in the crystals: residues 54 to 59 and 210 to 213 from the NH₂- and COOH-termini of pVHL, respectively, residues 50 to 57 of ElonginC, residues 107 to 118 of ElonginB, and residues 556 to 560 of HIF-1.

Isothermal titration calorimetry. For ITC experiments, the VBC complex was dialyzed against 20 mM bis-tris propane (pH 7.0), 75 mM NaCl, 5 mM DTT. The peptides were chemically synthesized, purified by reversed-phase chromatography, and dissolved in the buffer in which VBC was dialyzed. The ITC experiments were carried out in the Micro Calorimetry System (Microcal Inc.) at 25°C, by making 20 injections (5 l each) of the peptide solution into 1.3 ml of VBC. The concentrations of the protein and the peptides were measured by absorption at 280nm. The concentrations of the VBC employed were 0.051 mM for the measurements with the intact (D-E-M-L-A-Hyp-Y-I-P-M-D-D-D-F-Q-L-R-D) and the N segment (D-E-M-L-A-Hyp-Y-I-P-M) peptides, and 0.20 mM for the C segment peptide (D-D-D-F-Q-L-R-D). The extra aspartic acids on the termini of the peptides were included to improve their solubility. The concentration of the intact peptide was 1.9 mM, N segment 1.2 mM, and the C segment 22.0 mM. The curve fitting was done using MicroCal Origin software. The numbers shown are the averages and the standard deviations from 2 measurements for the intact peptide and the C segment, and from 3 measurements for the N segment. The binding curves and the parameters from the individual measurements are shown in the fig. S3.

pVHL far-Western blots. The peptides were synthesized on a derivatized cellulose filter using an Abimed Autospot 2000 robot according to the manufacturer's instructions (S6). The membrane containing the synthetic peptides was activated by methanol, and blocked with 4% milk in TBST (20 mM Tris-HCl, pH 7.6, 140 mM NaCl and 0.1% (v/v) Tween-20) for 2 h at room temperature. Purified VBC complex in 4% milk in TBST was then added to the blot at either 3 mg/ml or 0.3 mg/ml. After overnight incubation, the blot was washed three times with TBST followed by the addition of a 1:2000-diluted IG32 anti-pVHL antibody. After 2 h, the blot was washed and incubated with a 1:3000-dilution of goat-anti-mouse-AP antibody for 45 min. The blot was washed and developed with NBT and BCIP reagents. Results are shown in fig. S4.

Possible variability in HIF-1 bulge. Distance and steric considerations suggest that a bulge as short as two non-proline residues could still allow the N and C segments to bind pVHL, but owing to the curvature of the pVHL S4 strand, it would still correspond to an insertion relative to the register of backbone amide and carbonyl groups of the pVHL S4 and could not make continuous -sheet backbone hydrogen bonds. The bulge could also be substantially longer, although this may weaken binding due to the increased entropic cost of limiting the flexibility of the peptide upon pVHL binding.

Supplemental Figure 1. Stereo view of a F_o - F_c electron density difference map calculated using phases derived from the model after omitting the entire HIF-1 and subjecting the model to simulated annealing refinement from 3000°K using CNS, in order to remove model bias. Map was calculated at 1.85 Å resolution and contoured at 2.5 .

Medium version | Full size version

Supplemental Figure 2. Sequence alignment of the domain of pVHL from human, mouse, frog, fly and worm orthologues, highlighting residues identical in three or more of the five orthologues. Secondary structural elements of human pVHL are indicated (S1). The H4 helix, which is also a part of the domain, is not shown in the alignment. pVHL side chain and backbone groups that contact HIF-1 are indicated by green and orange dots, respectively. The histogram represents the 210 tumor-derived missense mutations in the pVHL residues 60 to 153 in the universal VHL-mutation database (S7).

Medium version | Full size version

Supplemental Figure 3. ITC titration curves and thermodynamic parameters

Medium version | Full size version

Supplemental Figure 4. pVHL far-Western blots with HIF-1 peptides. (A) Far-Western blot assay with HIF-1 peptides lacking the C segment. Successive NH₂- and COOH-terminal truncations of the peptide indicate that the bulge is not important for binding, that Ile⁵⁶⁶ is possibly the second most important side chain after Hyp⁵⁶⁴ and that Met⁵⁶¹, Leu⁵⁶² and Ala⁵⁶³ together make only a minor contribution to affinity. (B) To investigate the importance of the hydroxyproline group in specificity and affinity, we used the far-Western assay with a peptide library that contained all possible amino acids, except arginine, in place of the hydroxyproline in the context of a 12-mer N-segment peptide (558 D-L-E-M-L-A-Hyp-Y-I-P-M-D 569). While none of the amino acids restored wild-type binding, a long exposure of the blot showed that cysteine could weakly substitute for hydroxyproline. This is consistent with the mixed hydrophobic/polar environment in the binding site of Hyp⁵⁶⁴, as the cysteine sulfhydryl group can, in principle, provide both van der Waals contacts and hydrogen bonds, although it would not be isosteric with hydroxyproline.

Medium version | Full size version

Supplemental References

S1. C. E. Stebbins, W. G. Kaelin Jr., N. P. Pavletich, Science 284, 455 (1999).

S2. Z. Otwinowski, W. Minor, Methods Enzymol. 276, 307 (1997).

S3. CCP4, Acta Crystallogr. D 50, 760 (1994).

S4. T. A. Jones, J. Y. Zou, S. W. Cowan, M. Kjeldgaard, Acta Crystallogr. A 47, 110 (1991).

S5. A. T. Brunger et al., Acta Crystallogr. D 54, 905 (1998).

S6. R. Frank, H. Overwin, Methods Mol. Biol. 66, 149 (1996).

S7. C. Beroud et al., Nucleic Acids Res. 26, 256 (1998).