Evidence for a stereoelectronic effect in human oxygen sensing

Article abstract of DOI:10.1002/anie.200805427

How PHDs achieve specificity: trans-4-prolyl hydroxylation of the transcription factor HIF occurs with stereochemical retention. Substrate-analogue studies show how the von Hippel Lindau tumor suppressor protein (pVHL) and the oxygen-sensing hydroxylases (PHDs) achieve specificity for hydroxy prolyl/prolyl residues for the C4-exo/endo prolyl conformations, respectively.

Full text of DOI:10.1002/anie.200805427

Communications
DOI: 10.1002/anie.200805427
Conformational Analysis
Evidence for a Stereoelectronic Effect in Human Oxygen Sensing**
´
Christoph Loenarz, Jasmin Mecinovic, Rasheduzzaman Chowdhury, Luke A. McNeill,
Emily Flashman, and Christopher J. Schofield*
Human cells respond to limiting oxygen by
upregulation of the a,b-heterodimeric hypoxia
inducible transcription factor (HIF; for review
see Ref. [1]). Under limiting oxygen availability,
HIF increases expression of genes that work to
counteract hypoxic effects by enhancing blood-
vessel growth and red-blood-cell production. In
“normoxic” cells trans-4-prolyl-hydroxylation
of HIF-a enables its recognition by a ubiquitin
ligase, of which the von Hippel Lindau tumor
suppressor protein (pVHL) acts as a targeting
component. Thus, prolyl-hydroxylation marks
HIF for proteasomal degradation, and down-
regulates the expression of HIF target genes
(Figure 1).[1] HIF-a hydroxylation is catalyzed
by prolyl hydroxylase domain containing
enzymes (PHDs), which are FeII and 2-oxoglu-
tarate (2OG) dependent oxygenases.[2] The
Figure 1. The human HIF oxygen-sensing mechanism highlighting the role of HIF-a
absolute requirement of the PHDs for dioxygen prolyl hydroxylation.
enables them to act as oxygen sensors.[3] In
human cells there are two HIF hydroxylation
sites (in HIF-1a, these are Pro402 and Pro564) and three
PHD1-3 enzymes, of which the most important in normal
tissues is PHD2.[4] The selectivity of the interactions between
hydroxylated and non-hydroxylated HIF-a forms with pVHL
and the PHDs is central to the oxygen-sensing mechanism.
Herein we describe studies on the chemistry of these
interactions, which provide evidence for a role for the
stereoelectronic gauche effect in oxygen sensing.
tion (Figure 2).[5] Substitution with electronegative groups at
the C3 or C4 positions establishes a vicinal N-C-C-X arrange-
ment between the prolyl amide group and the electronegative
atom.[6] It is proposed that the “gauche effect” causes
electron-withdrawing trans-4 and cis-4 prolyl substituents to
effect a bias towards the C4-exo and C4-endo conformations,
respectively.[5] Thus, it is predicted that prolyl trans-4-hydrox-
ylation will bias the C4-endo/exo equilibrium to the exo form.
Indeed the C4-exo prolyl conformation present in trans-4-
hydroxyprolyl residues is crucial in maintaining the integrity
of the collagen triple helix.[5] Recently, Gorres et al. have
reported evidence that prolyl analogues with a C4-endo
preference are substrates of the collagen prolyl-4-hydroxy-
The pyrrolidine ring of prolyl residues can adopt C4-exo
and C4-endo conformations, with unsubstituted prolyl resi-
dues exhibiting a small preference for the C4-endo conforma-
´
[*] C. Loenarz, J. Mecinovic, R. Chowdhury, Dr. L. A. McNeill,
Dr. E. Flashman, Prof. C. J. Schofield
Chemistry Research Laboratory and the Oxford Centre for Integra-
tive Systems Biology, Department of Chemistry, University of Oxford
12 Mansfield Road, Oxford, OX1 3TA (UK)
Fax: (+44)1865-275-674
E-mail: christopher.schofield@chem.ox.ac.uk
Homepage: http://www.chem.ox.ac.uk/oc/cjschofield
[**] Supported by the BBSRC and The Wellcome Trust; Studienstiftung,
Dr.-Carl-Duisberg-Stiftung, and Rhodes Trust (C.L.); Newton-Abra-
ham Fund (J.M.); Commonwealth Scholarship Commission (R.C.).
We thank J. D. Webb and Prof. Dr. M. E. Maier (University of
Tꢀbingen) for assistance with [35S]-pVHL assays and help with
synthesis, respectively. Abbreviations: 2OG, 2-oxoglutarate; HIF,
hypoxia inducible factor; PHD, prolyl hydroxylase domain; pVHL,
von Hippel Lindau protein; ESI, electrospray ionization.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200805427.
Figure 2. Ring conformations of 4-substituted prolyl residues.
1784
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1784 –1787

Angewandte
Chemie
lase.[7] Herein we report our investigations on a role for the
gauche effect in the HIF system.
We initially synthesized HIF-1a556–574 peptides incorporat-
ing the Pro564 hydroxylation site with prolyl analogues that
are predicted to preferentially adopt C4-endo or C4-exo
conformations and assayed them for binding to, and as
substrates of human PHDs. Upon incubation with PHD2181–426
(tPHD2, a readily prepared truncated form of PHD2), a HIF-
1a556–574 peptide with trans-4-fluoroprolyl-564 residue was not
hydroxylated, but stimulated 2OG turnover (Supporting
Information, Figures S1, S2). In contrast, whilst both HIF
cis-4-hydroxyprolyl and cis-4-fluoroprolyl-564 analogues also
displayed high 2OG turnover, they underwent conversion
into a 4-oxoprolyl product, presumably by hydroxylation
followed by elimination of water or fluoride, respectively
À
(Figure 3). The “indirect” cleavage of a C F bond by PHDs is
notable since such enzyme-mediated processes are rare.[8]
Considering the stereoelectronic gauche effect, these results
imply favored hydroxylation of prolyl analogues with a
preference for the C4-endo, compared to a C4-exo, prolyl
conformation.
Figure 4. Non-denaturing ESI-MS analyses of tPHD2 in the presence
of equimolar FeII and HIF-1a peptides with prolyl-564 analogues;
peaks represent tPHD2·Fe (left) and tPHD2·Fe·HIF-a complexes
(right).
Oba et al. (Scheme 1).[10]
However, we were unable to
perform the saturation step
with the reported diastereo-
selectivities. Several catalysts
were investigated (Support-
ing Information, Table S1)
and
Crabtreeꢀs
catalyst
yielded approximately 96%
of the desired diastereomer 3
(Supporting
Information,
Figure S3). Incubation of a
HIF-1a556–574 peptide contain-
ing trans-4-D at the hydrox-
ylation site Pro564 with
tPHD2 gave hydroxylated
HIF-a with more than 95%
loss of the C4 deuterium
(mass increase of + 15 Da),
revealing that hydroxylation
occurs with retention of ste-
Figure 3. MALDI-TOF MS of HIF-1a peptides with Pro564 and cis-4-Flp564 analogues incubated with tPHD2.
DNPH=2,4-dinitrophenyl hydrazine; Flp=fluoroprolyl; R=2,4-dinitrophenyl.
We investigated the affinity of HIF substrates containing
the prolyl analogues for tPHD2. In support of the catalytic
assays, the affinity of HIF-1a556–574 Pro564 analogues with a
preference for the C4-endo prolyl conformation to tPHD2 was
found to be approximately five-fold stronger than for
analogues stabilizing the C4-exo conformation (by non-
denaturing ESI-MS analyses; see Figure 4). We propose that
the C4-endo prolyl conformation is required for productive
reochemistry (Figure 5; see the Supporting Information,
Figure S4 for high-resolution analyses). These results imply
that PHD catalysis proceeds by hydroxylation of HIF-1a
Pro564 in its C4-endo conformation and with retention of
stereochemistry. Upon hydroxylation, a bias to the C4-exo
conformation may aid in product release.
To investigate the generality of our proposals we carried
out incubations of HIF-1a395–413 peptides incorporating the N-
terminal Pro402 hydroxylation site, or prolyl-402 analogues,
with tPHD2 (Supporting Information, Figure S5). The results
support a similar binding mode of the substrate prolyl residue
at both the Pro402 and Pro564 HIF-1a hydroxylation sites.
Together with the work of Gorres et al.,[7] our results indicate
that a preference for the C4-endo conformation may be a
general property of prolyl-4-hydroxylases. Inspection of the
catalysis, that is, for reaction of the likely {FeIV O} inter-
=
mediate[9] with the C4-trans prolyl hydrogen atom.
To investigate the stereochemical course of trans-4-prolyl
hydroxylation of HIF-a, we prepared proline 4 appropriately
labeled with deuterium at C4, incorporated it into the Pro564
position of a HIF-1a556–574 peptide, and incubated the peptide
with tPHD2. Synthesis of 3 proceeded essentially according to
Angew. Chem. Int. Ed. 2009, 48, 1784 –1787
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1785

Communications
none of the cis-4-fluoro-/hydroxyprolyl or 4-oxoprolyl ana-
logues bound to pVHL within the detection limits of a pull-
down assay (Supporting Information, Figure S6);[11] signifi-
cantly neither did the trans-4-fluoroprolyl analogues. Non-
denaturing ESI-MS analyses also indicated that none of the
cis-4-fluoro-/hydroxyprolyl-564 analogues bound significantly
to VCB (Supporting Information, Figure S7).
To quantify the relative binding of the HIF peptides to
VCB, we carried out fluorescence-based assays.[13] Whilst
unmodified HIF-1a556–574 did not bind, the cis-4-hydroxyprolyl
analogue displayed a low affinity for pVHL (5.8 Æ 0.1% as
compared to trans-4-hydroxyprolyl); none of the other
analogues bound (Supporting Information, Figures S8, S9).
These results reveal the remarkable
Scheme 1. Synthesis of labeled proline; 1) 1.1 equiv SOCl2, EtOH, 16 h,
reflux; 2) 3 equiv Et3N, 1.2 equiv (Boc)2O, CH2Cl2, 17 h; 3) 1 equiv
PPh3, 1 equiv di-iPr azodicarboxylate, THF, 08C, then 1 equiv CH3I,
21 h; 4) 1.1 equiv PhSeNa (1.1 equiv Ph2Se2, EtOH, 1.1 equiv NaBH4,
16 h), 16 h, reflux; 5) 10 equiv H2O2, THF, 4 h, 08C; 6) 1 mol%
catalyst, CD3OD, 5 atm D2, 5 days; 7) 1m HCl, 3 h, reflux; 8) Dowex
50W-X8, lyophilization; 9) 1 equiv Et3N, H2O, 0.9 equiv Fmoc succini-
mide, MeCN, pH 9. Boc=tert-butyloxycarbonyl, Fmoc=(9H-fluoren-9-
ylmethoxy)carbonyl.
selectivity of the pVHL binding pocket
for HIF trans-4-hydroxyprolyl-402/564
and, since fluorine cannot substitute for
the trans-4-hydroxy group, demonstrate
that both hydrogen bonds from pVHL
to the hydroxy group of HIF trans-4-
hydroxyprolyl must contribute signifi-
cantly to the binding of HIF to pVHL
(Figure 5). Given the recent identifica-
tion of the reversibility of histone meth-
ylation, including by 2OG oxygenases
(for review see Ref. [2]), these results
raise the possibility that the signaling
effect of HIF hydroxylation may be
functionally ablated by oxidation to
the 4-oxoprolyl derivatives.
Overall our results reveal that cat-
alysis of HIF prolyl hydroxylation by
the oxygen-sensing human PHD1-3
Figure 5. Proposed role of conformational changes at HIF-1a Pro564/402 in molecular recog-
nition. Selected components shown only; when H*=D, loss of D was observed.
PHD’=PHD·FeII·2OG.
enzymes induces a conformational bias
from C4-endo to C4-exo (Figure 5) that is
an integral part of the hypoxic response
mechanism. Although the importance
of stereoelectronic effects in biology is
reactions catalyzed by human 2OG oxygenases reveals that
hydroxylation frequently occurs vicinal to an electronegative
atom. This result suggests possible involvement of stereo-
electronic effects in other 2OG oxygenase reactions, including
asparaginyl/aspartyl 3-hydroxylation, phytanoyl CoA 2-hy-
droxylation, lysyl 5-hydroxylation, and g-butyrobetaine 3-
hydroxylation.[2]
Given the apparently important role of the prolyl
conformation in PHD catalysis, we then investigated its role
in pVHL binding. The ability of a ternary pVHL/elongin C/B
protein complex (VCB) to discriminate between hydroxy-
lated and non-hydroxylated HIF-1a556–574 is remarkable (Kd =
33 nm and 34 mm, respectively).[11] In VCB·HIF crystal
structures,[11,12] the HIF trans-4-hydroxyprolyl-564 residue
adopts the C4-exo conformation; the hydroxy group is
positioned to make two hydrogen bonds (to pVHL Ser111
and His115, Figure 1). HIF-1a peptides with Pro564 and
Pro402 analogues containing C4 substituents known to bias
the ring to C4-endo or C4-exo conformations, as well as the 4-
oxoprolyl-564 derivative, were screened for pVHL binding.
Excepting the HIF trans-4-hydroxyprolyl-402/564 peptides,
established in carbohydrate chemistry with the anomeric
effect,[14] its role in post-translational modifications and
catalysis by other types of enzymes has not been widely
considered outside the context of collagen stabilization.[5]
Coupled with recent reports suggesting that post-translational
hydroxylation of intracellular proteins may be more wide-
spread than previously perceived,[15] our results suggest that
this effect may be involved in modulating other protein–
protein interactions.
Experimental Section
Methods Summary; see the Supporting Information for details and
abbreviations. Recombinant human PHD2181–426 (tPHD2) and VCB
complex were produced in E. coli and purified by affinity- and gel-
filtration chromatography. VCB was labeled with EuIII-cryptate
complex for FRET assays as reported.[13] Enzymatic activity was
analyzed by incubating tPHD2 with FeII, 2OG, and peptide and
assayed by 1-[14C]-2OG decarboxylation, DNPH derivatization,
MALDI-TOF, FT-ICR MS, and MS/MS analyses. Investigation of
reaction stereochemistry employed HIF-1a peptides prepared by
solid-phase peptide synthesis with proline analogues including 4. The
1786
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1784 –1787

Angewandte
Chemie
interaction of HIF-1a analogues with pVHL was assessed by IVTT
[6] W. Kim, K. I. Hardcastle, V. P. Conticello, Angew. Chem. 2006,
[35S]-pVHL capture assays, non-denaturing ESI-MS, and FRET.
118, 8321 – 8325; Angew. Chem. Int. Ed. 2006, 45, 8141 – 8145.
[7] K. L. Gorres, R. Edupuganti, G. R. Krow, R. T. Raines, Bio-
chemistry 2008, 47, 9447 – 9455.
[8] H. Deng, D. OꢀHagan, C. Schaffrath, Nat. Prod. Rep. 2004, 21,
773 – 784.
Received: November 6, 2008
Revised: December 3, 2008
Published online: January 29, 2009
[9] J. C. Price, E. W. Barr, B. Tirupati, J. M. Bollinger, Jr., C. Krebs,
Biochemistry 2003, 42, 7497 – 7508.
[10] M. Oba, T. Terauchi, A. Miyakawa, K. Nishiyama, Tetrahedron:
Asymmetry 1999, 10, 937 – 945.
[11] W. C. Hon, M. I. Wilson, K. Harlos, T. D. Claridge, C. J. Scho-
field, C. W. Pugh, P. H. Maxwell, P. J. Ratcliffe, D. I. Stuart, E. Y.
Jones, Nature 2002, 417, 975 – 978.
[12] J. H. Min, H. Yang, M. Ivan, F. Gertler, W. G. Kaelin, Jr., N. P.
Pavletich, Science 2002, 296, 1886 – 1889.
Keywords: conformational analysis · oxygen sensing ·
post-translational modifications · signal transduction
.
[1] G. L. Semenza, Curr. Opin. Cell Biol. 2001, 13, 167 – 171; C. J.
Schofield, P. J. Ratcliffe, Nat. Rev. Mol. Cell Biol. 2004, 5, 343 –
354.
[2] C. Loenarz, C. J. Schofield, Nat. Chem. Biol. 2008, 4, 152 – 156.
[3] L. A. McNeill, K. S. Hewitson, J. Gleadle, L. E. Horsfall, N. J.
Oldham, P. Maxwell, C. W. Pugh, P. J. Ratcliffe, C. J. Schofield,
Bioorg. Med. Chem. Lett. 2002, 12, 1547 – 1550.
[4] E. Berra, E. Benizri, A. Ginouves, V. Volmat, D. Roux, J.
Pouyssegur, EMBO J. 2003, 22, 4082 – 4090.
[13] J. H. Dao, R. J. M. Kurzeja, J. M. Morachis, H. Veith, J. Lewis, V.
Yu, C. M. Tegley, P. Tagari, Anal. Biochem. 2009, 384, 213 – 223.
[14] P. Deslongchamps, Stereoelectronic Effects in Organic Chemistry,
Pergamon, Oxford, 1983.
[15] M. E. Cockman et al., Proc. Natl. Acad. Sci. USA 2006, 103,
14767 – 14772; see the Supporting Information.
[5] J. A. Hodges, R. T. Raines, J. Am. Chem. Soc. 2005, 127, 15923 –
15932.
Angew. Chem. Int. Ed. 2009, 48, 1784 –1787
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1787