International Journal of

Molecular Sciences

Article
Structural Characterization of Hypoxia Inducible Factor
α—Prolyl Hydroxylase Domain 2 Interaction through
MD Simulations
Giorgia F. Camagni † , Giovanni Minervini † and Silvio C. E. Tosatto *

Department of Biomedical Sciences, University of Padua, 35121 Padova, Italy

* Correspondence: [email protected]
† These authors contributed equally to this work.

Abstract: The Prolyl Hydroxylases (PHDs) are an enzymatic family that regulates cell oxygen-sensing.
PHDs hydroxylate hypoxia-inducible transcription factors α (HIFs-α) driving their proteasomal
degradation. Hypoxia inhibits PHDs activity, inducing HIFs-α stabilization and cell adaptation to
hypoxia. As a hallmark of cancer, hypoxia promotes neo-angiogenesis and cell proliferation. PHD
isoforms are thought to have a variable impact on tumor progression. All isoforms hydroxylate
HIF-α (HIF-1,2,3α) with different affinities. However, what determines these differences and how
they pair with tumor growth is poorly understood. Here, molecular dynamics simulations were
used to characterize the PHD2 binding properties in complexes with HIF-1α and HIF-2α. In parallel,
conservation analysis and binding free energy calculations were performed to better understand
PHD2 substrate affinity. Our data suggest a direct association between the PHD2 C-terminus and
HIF-2α that is not observed in the PHD2/HIF-1α complex. Furthermore, our results indicate that
phosphorylation of a PHD2 residue, Thr405, causes a variation in binding energy, despite the fact
that this PTM has only a limited structural impact on PHD2/HIFs-α complexes. Collectively, our
findings suggest that the PHD2 C-terminus may act as a molecular regulator of PHD’s activity.

Keywords: molecular dynamics simulation; hypoxia; hypoxia-inducible factor; HIF-prolyl

hydroxylases; VHL
Citation: Camagni, G.F.; Minervini,
G.; Tosatto, S.C.E. Structural
Characterization of Hypoxia
Inducible Factor α—Prolyl
Hydroxylase Domain 2 Interaction
1. Introduction
through MD Simulations. Int. J. Mol. Hypoxia is a major hallmark of tumor growth. The mechanism of adaptation to
Sci. 2023, 24, 4710. https://doi.org/ hypoxia at the cellular level is mediated by the hypoxia-inducible factor (HIF) family, a tran-
10.3390/ijms24054710 scription factor consisting of an oxygen-sensitive alpha subunit (HIF-1α and HIF-2α being
the most studied isoforms) and a constitutively expressed beta subunit (HIF-β/ARNT) [1].
Academic Editor: Renata Tisi
Its activity is regulated by Prolyl Hydroxylase Domain-Containing Proteins (PHDs) be-
Received: 15 December 2022 longing to the 2-oxoglutarate-dependent dioxygenases superfamily. The enzymatic activity
Revised: 23 February 2023 of these enzymes depends on a ferrous iron cofactor, a co-substrate, i.e., 2-oxoglutarate
Accepted: 25 February 2023 (2-OG), and molecular oxygen [2]. In normoxic conditions, PHDs hydroxylate specific
Published: 1 March 2023 proline residues (Pro402 and Pro564 in HIF-1α, Pro405, and Pro531 in HIF-2α) located in the
so-called LxxLAP motifs of the HIFs-α family [3–6]. Hydroxylation allows the subsequent
interaction with the von Hippel–Lindau tumor suppressor protein (pVHL), a member of
Copyright: © 2023 by the authors.
the ubiquitin E3 ligase complex termed VCB, which includes Elongin-B, Elongin-C, and
Licensee MDPI, Basel, Switzerland.
Cullin-2p [7,8]. This interaction leads to the ubiquitination and proteasome-mediated
This article is an open access article degradation of hypoxia-inducible factors α [9,10]. Furthermore, pVHL is the main actor
distributed under the terms and in von Hippel–Lindau disease (VHL) [11], a human hereditary predisposition to develop
conditions of the Creative Commons cancer caused by mutation of the homonymous gene. Typical manifestations of the diseases
Attribution (CC BY) license (https:// include clear-cell renal cell carcinoma (ccRCC), retinal- and cerebellar-hemangioblastoma,
creativecommons.org/licenses/by/ pheochromocytoma (PCC), and pancreatic neuroendocrine tumors, in addition to pancre-
4.0/). atic and renal cysts [12–15]. Hypoxic conditions reduce PHD activity allowing HIFs-α to

Int. J. Mol. Sci. 2023, 24, 4710. https://doi.org/10.3390/ijms24054710 https://www.mdpi.com/journal/ijms

Int. J. Mol. Sci. 2023, 24, 4710 2 of 15

escape pVHL recognition. HIFs-α accumulate in the cytosol and translocate to the nucleus,
where they dimerize with HIF-1β to regulate the transcription of multiple genes involved
in hypoxia adaptive response, including the vascular endothelial growth factor (VEGF),
erythropoietin (EPO), pyruvate dehydrogenase kinase-1 (PDK1), and glucose transporter-1
(GLUT-1) [16,17]. Although the PHDs are mostly inhibited in hypoxia, their activity is
still observed at low oxygen levels, highlighting their sensitivity to changes in oxygen
availability and their role as cellular oxygen sensors [18,19]. The PHDs family consists
of three canonical members: PHD1 (EglN2), PHD2 (EglN1), PHD3 (EglN3), and a more
recently discovered PHD4-TM (EglN4). All share a dioxygenase domain at the C-terminal,
while the N-terminal is less conserved [20]. They show specific subcellular localizations
and tissue-specific expressions. Indeed, PHD1 is predominantly expressed in the nucleus,
PHD2 in the cytoplasm, and PHD3 in both. All these isoforms are ubiquitously expressed in
human tissues; however, PHD2 is reported to be more abundant than others [21]. PHD1 is
predominantly found in the testes, brain, kidney, heart, and liver, whereas PHD3 is mostly
expressed in the cardiac tissue [21,22]. All members of this enzyme family hydroxylate
the conserved proline residues present in HIFs-α family members; however, they present
different affinities. PHD2 appears to have a greater affinity for HIF-1α under normoxic
conditions, while PHD1 and PHD3 are proposed to predominantly contribute to HIF-2α
regulation [23–25]. The cell oxygen sensing system is altered in many tumors, which prefer
glycolytic metabolism despite being in the presence of oxygen (the Warburg effect). This
results in the inactivation of prolyl hydroxylases, stabilization of HIFs-α, and subsequently,
angiogenesis, proliferation, and cell survival to occur. Since PHDs directly regulate HIFs-α
activity, they can be considered a central regulator of tumor development. However, their
behavior in the context of tumors remains controversial. It has been observed that PHD
isoforms have a variable and cell-dependent impact on tumor progression. Their inhibition
can either promote or inhibit tumor proliferation. For instance, PHD3 inhibits colon and
gastric cancer growth while promoting it in the ccRCC and maintaining high levels of
HIF-2α. Again, overexpression of PHD2 inhibits liver cancer growth. On the other hand, its
inhibition reduces the growth of osteosarcoma [24,25]. What determines the pro- and anti-
tumor functions of each isoform, as well as their different substrate affinities is still poorly
understood. Here, we used molecular dynamics simulations to characterize the PHD2
substrate specificity in binding HIF-1α and HIF-2α by identifying specific intermolecular
interactions of each substrate. We also simulated the phospho-Thr405 (TPO) located on the
PHD2 C-terminus to investigate its role in the binding process. Finally, we calculated the
binding free energy to better understand the substrate affinity.

2. Results
2.1. Homology Modeling
We started our investigation by generating a tridimensional structure of the PHD2/HIF-
2α complex by homology modeling. To this end, the PHD2/HIF-1α complex crystallized
structure (PDB ID: 6YW3) has been used as a template. The experimental structure of HIF-
1α in complex with PHD2 covers 17 amino acids (558–574), a linear motif corresponding
to its C-terminal oxygen-dependent degradation domain (CODD), and also includes the
conserved LxxLAP sequence motif, where the P indicates the hydroxyl acceptor proline
(Figure 1).
As there is insufficient data to indicate how HIF-2α arranges itself in the enzymatic
binding pocket, we hypothesize that the position of the proline subjected to hydroxylation
(HIF-1α Pro564 and HIF-2α Pro531) should be preserved. The alignment and the resulting
model are shown in Figure 2.
Int. J. Mol. Sci. 2023, 24, x FOR PEER REVIEW 3 of 18
Int. J. Mol. Sci. 2023, 24, 4710 3 of 15

Figure 1. Overview
Figure 1. OverviewofofPHD2PHD2 structure.
structure. (A) (A) A cartoon
A cartoon representation
representation of the enzyme
of the PHD2 PHD2 enzyme in
in complex
complex with its targets. The HIF-1α peptide is represented in orange, while purple is for
with its targets. The HIF-1α peptide is represented in orange, while purple is for the HIF-2α peptide. the HIF-
2α
(B)peptide. (B) A view
A zoomed-in zoomed-in
of the view
PHD2ofcatalytic
the PHD2 catalytic
site. site. Key
Key residues forresidues
substrateforbinding
substrate
and binding and
enzymatic
enzymatic activity are presented as sticks. (C) A mesh view highlighting the PHD2 binding site in
activity are presented as sticks. (C) A mesh view highlighting the PHD2 binding site in complex with
complex with the HIF-s substrates and co-substrate α-ketoglutarate (AKG).
the HIF-s substrates and co-substrate α-ketoglutarate (AKG).
As there is
2.2. Molecular insufficient
Dynamics data to and
Simulations indicate how HIF-2α
Interaction Analysisarranges itself in the enzymatic
binding
2.2.1. PHD2/HIF-1α and PHD2/HIF-2α Complexesthe proline subjected to hydroxylation
pocket, we hypothesize that the position of
(HIF-1α Pro564 and HIF-2α Pro531) should be preserved. The alignment and the resulting
The crystal structure of PHD2 is composed of the domain containing the catalytic site
model are shown in Figure 2.
(185–407) and two disordered regions, i.e., the β2β3-loop (237–254), which is involved in
the binding process, and the C-terminus (400–407). In order to characterize the substrate
specificity of this isoform and identify specific inter-molecule interactions, 1 µs long molec-
ular dynamics (MD) simulations of the complexes formed by PHD2 and HIF-1α/HIF-2α
were carried out (Figure 3). The root-mean-square deviation (RMSD) and root-mean-square
fluctuation (RMSF) plots of the PHD2/HIF-1α complex indicate that the systems remain
stable for the entire simulation time with only moderate fluctuations (Figure 3A). The
regions showing the greatest fluctuations are located at residues 237–254 and at residues
400–407, corresponding to the β2β3-loop and the C-terminus, respectively (Figure 3A).
Differently, the PHD2/HIF-2α complex shows an important entropic effect in all simulation
runs, with RMSD values reaching ~6 Å (Figure 3B). As shown in Figure 3B, regions with
significant fluctuations are consistently observed in the β2β3-loop and C-terminal regions.
Such behavior in both systems is expected, as these are disordered regions characterized
by a high degree of conformational freedom. In detail, in the first simulation (orange), the
complex is always stable and assumes a conformation in which the C-terminal region inter-
acts with HIF-2α (Figure 4). The same conformation is observed in the second simulation
run (blue) between 100 and 250 ns. Moreover, the system is very stable even in the last
300 ns of simulation, in which the PHD2 C-terminus appears to assume a mainly alpha
secondary structure (Figure 4).
 Int. J. Mol. Sci. 2023, 24, x FOR PEER REVIEW 4 of 18
Int. J. Mol. Sci. 2023, 24, 4710 4 of 15

Figure 2.
Figure 2. Sequence
Sequencealignment
alignmentand andhomology
homologymodeling
modelingofofHIF-2α.
HIF-2α.(A)(A)Sequence
Sequence alignment
alignmentbetween
between
the 17
the 17 amino
amino acid
acidresidues
residuesof ofHIF-1α
HIF-1α(558–574)
(558–574)and
andHIF-2α
HIF-2α (526–542).
(526–542). In In
red, thethe
red, hydroxyl
hydroxylacceptor
acceptor
proline. LxxLAP
proline. LxxLAP motifs
motifs are
are shown
shown inin the
the light
light blue
blue box.
box. In
In yellow,
yellow, the
the presence
presence ofof glycine
glycine in
in HIF-
HIF-2α
2α leads
leads to the
to the formation
formation of aofgap
a gap in the
in the HIF-1α
HIF-1α sequence.
sequence. (B)(B) The
The leftimage
left imageshows
showsthethesurfaces
surfacesofofthe
the PHD2 complex (teal), with the HIF-1α (orange)/HIF-2α (purple) superposed. The right box
PHD2 complex (teal), with the HIF-1α (orange)/HIF-2α (purple) superposed. The right box shows
shows the catalytic site and the conserved HIF-2α Pro531 position.
the catalytic site and the conserved HIF-2α Pro531 position.
2.2. Molecular
Based onDynamics Simulations
these findings, and Interactionthat
we hypothesized Analysis
the C-terminus may play a role in the
2.2.1. PHD2/HIF-1α
binding and PHD2/HIF-2α
process by stabilizing Complexes
the substrate in the active site. Indeed, we observed a RMSD
shift The crystal structure of PHD2 is composed of
depending on the conformation assumed bythe
thedomain containing
C-terminus, the value
with the catalytic site
reaching
5(185–407)
Å whenand thetwo disordered regions, i.e., the β2β3-loop (237–254), which is involved
C-terminus directly interacts with HIF-2α. In contrast, a lower in
RMSD
the binding
value of 3.5 process, and thewhen
Å is registered C-terminus (400–407). assumes
the C-terminus In order to characterize
a mainly alphathe substrate
conformation.
specificity
These of this
findings isoform
suggest and
that theidentify
bindingspecific inter-molecule interactions,
site closing/opening due to C-terminus 1 µs long
steric
molecular may
hindrance dynamics
increase (MD)the simulations
energy content of the complexes
of the complex.formed
A certainby degree
PHD2 of and HIF-
instability,
1α/HIF-2αwas
however, were carried
also out (Figure
registered in the3). Thereplica,
third root-mean-square deviation (RMSD)
where the C-terminus assumed andmultiple
root-
mean-square fluctuation
conformations during the(RMSF) plots of thetime.
entire simulation PHD2/HIF-1α
At 580 ns in complex indicate
particular, that the a
we observed
systems spike
sudden remain stable
in the RMSDfor value
the entire
that issimulation
promotedtime by the with only moderate
breaking of van derfluctuations
Waals (VDW)
(Figure 3A). The
interactions regions
between theshowing the greatest
PHD2 residue Trp258 fluctuations
and HIF-1α arePro534.
locatedIn at order
residues 237–254
to investigate
and at residues 400–407, corresponding to the β2β3-loop and the C-terminus, respectively
the molecular details behind the substrate specificity reported for PHDs, the specific and
(Figure 3A). Differently,
non-specific intermolecular the PHD2/HIF-2α
interactions of complex
PHD2shows an important
in complex entropic
with HIF-1α effect
and in
HIF-2α
all simulation
were runs, with
also analyzed. BothRMSD
of thesevalues reaching
substrates ~6 Å (Figure
consist 3B). As
of 17 amino shown
acids, in Figure
spanning 3B,
residues
regions with
558–574 significant
and 526–542 for fluctuations
HIF-1α and HIF-2α,are consistently observed
respectively in 2A).
(Figure the β2β3-loop
Our analysis andshows
C-
terminal
that regions.
two bonds Such behavior
formed in both systems
by Asp536-Arg396 is expected, as
and Glu538-Arg396 arethese are disordered
relevant in stabilizing
regions in
HIF-2α characterized
the binding by a high
pocket. Thedegree
sameof conformational
Arg also interacts freedom.
with Pro567In detail,
of HIF-1α;in the first
however,
simulation
this (orange),
interaction appears the not
complex
to be is always
shared stablereplicas.
among and assumes a conformation
Furthermore, HIF-1αinformswhich two
the C-terminal
very region interacts
stable H-bonds with PHD2; with HIF-2α (Figure 4).
Leu562-Tyr310 andThePro564-Arg322.
same conformation is observed
In both systems,
in the second
Lys297 residue simulation
seems torun play(blue)
a rolebetween 100 and
in stabilizing the250 ns. Moreover,
substrate, the system
especially in theiscomplex
very
stable even in the last 300 ns of simulation, in which the PHD2
containing HIF-2α. Based on our simulations, PHD2 Lys297 forms ionic bonds with HIF-2α C-terminus appears to
assume aAsp536,
residues mainly alpha
Glu538, secondary
and Asp539.structure (FigureLys297
Similarly, 4). engages electrostatic interactions
with Asp570 and Asp569 when complexed with HIF-1α. The role of these negatively
charged residues in driving the binding with PHD2 is also supported by their conservation
(Figure 2A). In the PHD2/HIF-1α complex, the ionic bonds that Asp571 establishes with
Arg396 and Lys400 are therefore more relevant. In two simulation runs of the PHD2/HIF-
2α complex, we observed a single ionic bond that seemed to stabilize the direct interaction
between the PHD2 C-terminus and the substrate. In particular, Lys402 residue interacts
Int. J. Mol. Sci. 2023, 24, 4710 5 of 15

with Asp536 throughout the entire first simulation, while in the second simulation run,
Lys402 interacts with Asp539 for ~150 ns. Contrarily, the ionic bond formed by the pair
Glu538-Lys262, observed only in the last 300 ns of the second simulation run, suggests
that this specific interaction is essential to stabilize a complex conformation, prompting the
PHD2 C-terminus to increase its secondary structure content (Figure 4). Multiple constant
VDW interactions were also identified in all simulations. Among them, we find that Phe391
interacts with Met535 of HIF-2α and Asp571 of HIF-1α. The same isoleucine (namely Ile533
in HIF-2α and Ile566 in HIF-1α) interacts with Arg322 and Thr296. All these interactions
are more frequently observed in the HIF-1α complex than in the HIF-2α complex, where
the binding to Trp389 is more stable than the others. The Trp258 residue binding Pro534 of
HIF-2α and Pro567 of HIF-1α turns out to be an important interaction for both substrates.
These two interactions remain stable for the entire simulation time and are shared among
all replicas. Their breaking is also decisive in driving the conformational change occurring
at 580 ns in the third replica of the PHD2/HIF-2α complex. A conserved phenylalanine
residue of HIF-1α and HIF-2α (i.e., Phe572 and Phe540, respectively), establishes multiple
VDW interactions relevant for the stabilization of both substrates in the catalytic pocket.
Interestingly, it interacts with different residues depending on the substrate involved, as
its side chain assumes an opposite conformation in the two complexes. In particular, it
interacts with Arg295 in the PHD2/HIF-1α, while with PHD2/HIF-2α it interacts with
residues forming the fourth PHD2 α-helix, i.e., Arg396, Ala399, and Lys400. Additionally,
Leu574, the last residue included in the HIF-1α CODD, forms stable VDW interactions
with Asp277, Ile280, and Asn293. However, due to a difference in the amino acid sequence,
these bonds are not observed in the complex with HIF-2α (Figure 2A). Similarly, the pair,
Ala563-Pro317, is also PHD2/HIF-1α complex-specific. A major difference between the two
substrates concerns the interaction involving the proline residue targeted for hydroxylation.
Indeed, the HIF-1α Pro564 forms a stable VDW with His313, an amino acid belonging to the
catalytic triad in iron coordination, highlighting its importance in maintaining the catalytic
site. Instead, the HIF-2α Pro531 is stabilized by interacting with the β2β3-loop, particularly
with Val241. It also interacts with His313; however, this specific contact appears to be less
stable and was observed in just one simulation. These findings explain why, although the
HIF-2α N-terminus appears to have a higher conformational flexibility, the proline residue
is kept in the binding pocket in the correct position to be hydroxylated. Finally, HIF-2α
Phe540 forms a π-π stack with PHD2 Tyr403. This interaction is shared among replicas,
suggesting that it is relevant to stabilizing the complex. A similar π-π stack is also observed
in simulations of the PHD2/HIF-1α complex (i.e., the pair Tyr565-Trp258), however its
frequency is lower and inconstant among runs. All interactions are listed in Table 1.

2.2.2. Phosphorylation of PHD2-Thr405

Considering our results, which indicate the PHD2 C-terminus as a putative relevant
actor in modulating PHD2 substrate specificity, we wondered whether post-translational
modification of residues in this region may activate a further layer of regulation. The
PhosphoSitePlus [26] (www.phosphosite.org, accessed on 15 August 2022) database reports
that Thr405 located in the C-terminus is a phosphorylation site found in leukemia cells. To
investigate if this modification may have an impact on the PHD2/HIFs-α binding, we simu-
lated the phosphorylation of Thr405 (TPO) and evaluated its impact on the complexes. The
phosphorylated PHD2/HIF-1α system exhibits a general trajectory dynamic resembling
data obtained from the unmodified complex, also sharing a comparable degree of flexibility
for both the disordered regions forming the β2β3-loop and the C-terminus (Figure 5A).
This finding suggests that phospho-Thr405 does not affect the general stability of the com-
plex, as also indicated by the radius of gyration (Rg) profile, which remains stable in the
range of 17–18 Å in all simulations. Similarly, the modified PHD2 complexed with HIF-2α
shows a dynamic behavior comparable to that obtained from other simulations without
phosphorylation. Indeed, RMSD oscillations are similar, mainly due to the pronounced
flexibility of the β2β3-loop and C-terminus (Figure 5A). Measurement of Rg variation also
Int. J. Mol. Sci. 2023, 24, 4710 6 of 15

showed a substantial conservation of secondary structure; we then excluded any possible

local or long-distance influence of the phosphorylated C-terminus on the PHD2 structure.
Additionally, in these runs, the PHD C-terminus is the protein region showing the great-
Int. J. Mol. Sci. 2023, 24, x FOR PEER REVIEW 5 of 18
est conformational freedom, which influences protein stability. Its phosphorylation state,
however, appears to have only a modest or null effect on PHD2/HIFs-α.

Figure 3.
Figure 3. RMSD
RMSD and
and RMSF
RMSF plots
plotscalculated
calculatedononPHD2/HIF-1α
PHD2/HIF-1α and and PHD2/HIF-2
PHD2/HIF-2 α α complexes C-alpha
complexes C-alpha
over time. (A) RMSD (left) and RMSF (right) of the PHD2/HIF-1α complex. The
over time. (A) RMSD (left) and RMSF (right) of the PHD2/HIF-1α complex. The four simulations four simulations
that were
that were run
run are
are shown
shown inin different
different colors:
colors: the
the first
first (orange),
(orange), the
the second
second (blue),
(blue), the
the third
third (turquoise),
(turquoise),
and the fourth (green). (B) RMSD (left) and RMSF (right) plots of the PHD2/HIF-2α complex. The
and the fourth (green). (B) RMSD (left) and RMSF (right) plots of the PHD2/HIF-2α complex. The
first simulation is indicated in orange, the second in blue, and the third in turquoise.
first simulation is indicated in orange, the second in blue, and the third in turquoise.

Table 1. Conserved and specific interactions of PHD2/HIF-1α and PHD2/HIF-2α complexes with a
frequency cutoff < 20%. Gray is for conserved interactions, while HIF-1α specific interactions are
marked in orange. HIF-2α specific interactions are highlighted in purple.

PHD2 HIF-1α HIF-2α Interaction

Residue Residue Residue Type
Asp536 Run
Arg396 Pro567 H-Bond
Glu538 1◦ –2◦
Tyr310 Leu562 H-Bond
Arg322 Pro564 H-Bond
Asp536
Asp569
Lys297 Glu538 Ionic Bond
Asp570
Asp539
Arg396 Run
Asp571 Asp536 Ionic Bond
Lys400 3◦
Asp536 Run 1◦ –2◦
Lys402 Ionic Bond
Asp539 (C-ter closed conformation)
Int. J. Mol. Sci. 2023, 24, 4710 7 of 15

Int. J. Mol. Sci. 2023, 24, x FOR PEER REVIEW 5 of 18

Table 1. Cont.

PHD2 HIF-1α HIF-2α Interaction

Residue Residue Residue Type
Lys262 Glu538 Ionic Bond Run 2◦ –3◦
Run
Phe391 Asp571 Met535 VDW
1◦ –3◦
Thr296
Arg322 Ile566 Ile533 VDW
Trp389
Trp258 Pro567 Pro534 VDW
Arg295 Phe572 VDW
Arg396
Ala399 Phe540 VDW
Lys400
Run 1◦
Lys404 Phe540 VDW
(C-ter closed conformation)
Run 2◦
Val401 Phe540 VDW
(C-ter closed conformation)
Gln239 Leu562 Leu529
Leu240 Ala563 Ala530 VDW
Val241 Tyr565 Tyr532
Asp277
Ile280 Leu574 VDW
Asn293
Pro317 Ala563 Ala530 VDW Run 1◦ –2◦
Figure
His3133. RMSD and Pro564
RMSF plots calculated on PHD2/HIF-1α
Pro531 VDWand PHD2/HIF-2 α complexes C-alpha Run 2◦
over time. (A) RMSD
Val241 (left) and RMSF Pro531
Pro564 (right) of the PHD2/HIF-1α
VDW complex.
RunThe ◦
1 –2four simulations
◦
that were run are shown in different colors:
Tyr403 the first (orange),
Phe540 the second (blue), the third
π-π stack Run(turquoise),
1◦ –2◦
and the
Trp258 fourth (green). (B)
Tyr565 RMSD (left) and
Tyr532 RMSF (right) plots
π-π stackof the PHD2/HIF-2α
Run 2 ◦ complex. The3◦
Run
first simulation is indicated in orange, the second in blue, and the third in turquoise.

Figure 4. PHD2 C-terminus conformations. The C-terminus interacts with HIF-2α during the entire
first run (orange), and in the second run (blue) between 100 and 250 ns. Further, the C-terminus
appears to assume a mainly alpha secondary structure during the last 300 ns of the second run (blue).
Int.
Int. J.J. Mol.
Mol. Sci.
Sci. 2023,
2023, 24,
24, x4710
FOR PEER REVIEW 9 8of
of 18
15

Figure 5. Comparison
Figure 5. ComparisonofofRMSD,
RMSD, RMSF,
RMSF, andand Rg plots
Rg plots of PHD2/HIFs-a
of PHD2/HIFs-a complexes
complexes with Thr405-P
with Thr405-P (TPO).
(TPO). (A) PHD2/HIF-1α
(A) PHD2/HIF-1α + TPO (orange)
+ TPO (orange) and PHD2/HIF-2α
and PHD2/HIF-2α + TPO (purple).
+ TPO (purple). (B) Comparison
(B) Comparison of the
of the gyration
gyration radius between all systems of the PHD2/HIF-1α complex: first simulation (orange), second
radius between all systems of the PHD2/HIF-1α complex: first simulation (orange), second run (blue),
run (blue), third (turquoise), fourth (green), and simulation with phosphorylation (purple). (C)
third (turquoise), fourth (green), and simulation with phosphorylation (purple). (C) Comparison of
Comparison of gyration radii among all systems of the PHD2/HIF-2α complex: first simulation
gyration radii among all systems of the PHD2/HIF-2α complex: first simulation (orange), second
(orange), second simulation (blue), third simulation (turquoise), and simulation with
simulation (blue),(purple).
phosphorylation third simulation (turquoise), and simulation with phosphorylation (purple).

Interestingly, we observed some change in the residue-residue interactions, indicating

Interestingly, we observed some change in the residue-residue interactions,
that phopho-Thr405 may retain a possible regulative role in modulating PHD2 substrate
indicating that phopho-Thr405 may retain a possible regulative role in modulating PHD2
specificity. Our simulations also show that Hys313 within the PHD2 catalytic triad binds
substrate specificity. Our simulations also show that Hys313 within the PHD2 catalytic
to Ala530 with a more stable interaction than that observed in systems without phos-
triad binds to Ala530 with a more stable interaction than that observed in systems without
phorylation. Furthermore, during the entire simulation time, the direct interaction of the
phosphorylation. Furthermore, during the entire simulation time, the direct interaction of
C-terminal with HIF-2α is apparently lost, with HIF-2α engaging in a novel interaction with
the
the C-terminal
β2β3-loop with
via aHIF-2α
H-bondisformed
apparently lost,
by the with
pair HIF-2α engaging
Gly406-Lys244 in a 6).
(Figure novel interaction
Although this
with the β2β3-loop via a H-bond formed by the pair Gly406-Lys244 (Figure
bond does not persist for the whole simulation, it suggests that when PHD2 interacts 6). Although
with
this bondthe
HIF-2α, does not persistmay
C-terminus for assume
the whole simulation,
a dynamic it suggests
behavior, that itwhen
leading PHD2
to close theinteracts
binding
with HIF-2α, the C-terminus may assume a dynamic behavior, leading it
pocket. This data agrees with the idea of a PHD2 C-terminus involved in modulating PHD2 to close the
binding pocket. This
substrate specificity. data agrees with the idea of a PHD2 C-terminus involved in
modulating PHD2 substrate specificity.
2.3. Cluster Analysis
We then performed a RMSD-based structural hierarchical and state clustering analysis
to extract central conformational states from all the MD simulation trajectories (Table 2). As
expected, the conformation analysis shows that PHD2 tends to explore fewer conformations
when it is in complex with HIF-1α, i.e., an experimental 3D structure, indicating a higher
stability of this system with respect to the PHD2/HIF-2α complex (obtained from molecular
modeling). Phosphorylation of Thr405 (TPO) seems to further stabilize the PHD2/HIF-1α
as the number of clusters recapitulating the entire trajectory of this system is lower than
the average of those with no TPO. In contrast, TPO seems not to introduce considerable
conformational changes in the PHD2/HIF-2α complex.
Int. J.J. Mol.
Int. Mol. Sci.
Sci.2023, 24,x4710
2023,24, FOR PEER REVIEW 10 9ofof 18
15

Figure
Figure 6.
6. β2β3-loop
β2β3-loop andand C-terminus
C-terminus H-bond
H-bond interaction.
interaction. PHD2
PHD2(teal)
(teal) and
and HIF-2α
HIF-2α (purple)
(purple) are
are
represented in cartoon style. Lys244 and Gly406 residues are represented by licorice and H-bonds
represented in cartoon style. Lys244 and Gly406 residues are represented by licorice and H-bonds
(turquoise) by a stick.
(turquoise) by a stick.

2.3.
TableCluster
2. TotalAnalysis
cluster numbers of PHD2/HIF-1α, PHD2/HIF-2α, PHD2/HIF-2α with Thr405-P (TPO),
We then performed
and PHD2/HIF-2α a RMSD-based
with Thr405-P (TPO). structural hierarchical and state clustering
analysis to extract central conformational states from all the MD simulation trajectories
MD
(Table 2). As expected, the conformation PHD2/HIF-1α
analysis shows that PHD2 tends toPHD2/HIF-2α
explore fewer
PHD2/HIF-1α PHD2/HIF-2α
Simulations (TPO) (TPO)
conformations when it is in complex with HIF-1α, i.e., an experimental 3D structure,
◦
indicating1 a higher stability 9 of this system25with respect to the 12 PHD2/HIF-2α43complex
2◦ 25 30
(obtained◦ from molecular modeling). Phosphorylation of Thr405 (TPO) seems to further
3 19 73
stabilize4the
◦ PHD2/HIF-1α15 as the number of clusters recapitulating the entire trajectory of
this system is lower than the average of those with no TPO. In contrast, TPO seems not to
introduce considerable conformational changes in the PHD2/HIF-2α complex.
Then we selected the first six most populated clusters for each trajectory and extracted
the most
Table representative
2. Total conformations
cluster numbers (Figure
of PHD2/HIF-1α, 7). The regions
PHD2/HIF-2α, showing
PHD2/HIF-2α withsignificant con-
Thr405-P (TPO),
and PHD2/HIF-2α
formational withinThr405-P
change (TPO). are the β2β3-loop and C-terminus. This flexibility
both complexes
can be interpreted as a consequence of the disorder content in these specific portions
of PHD2 MD and is in agreement PHD2/HIF-1α PHD2/HIF-2α
PHD2/HIF-1α PHD2/HIF-2α
with what is observed with the RMSD and RMSF inspec-
Simulations (TPO)
tions. Significant changes are also observed at the substrate level, in particular (TPO)for the
1°
PHD2/HIF-2α 9
complex, where 25
one representative 12 describes the direct
conformation 43 inter-
2°
action between the protein25C-terminus and the30 substrate, lending support to its presumed
3°
role in substrate 19
discrimination. 73
4° 15

Then we selected the first six most populated clusters for each trajectory and
extracted the most representative conformations (Figure 7). The regions showing
significant conformational change in both complexes are the β2β3-loop and C-terminus.
This flexibility can be interpreted as a consequence of the disorder content in these specific
portions of PHD2 and is in agreement with what is observed with the RMSD and RMSF
 Int. J. Mol. Sci. 2023, 24, x FOR PEER REVIEW 11 of 18

inspections. Significant changes are also observed at the substrate level, in particular for
the PHD2/HIF-2α complex, where one representative conformation describes the direct
Int. J. Mol. Sci. 2023, 24, 4710 interaction between the protein C-terminus and the substrate, lending support to10its of 15
presumed role in substrate discrimination.

Figure
Figure 7. 7.Superimpositions
Superimpositionsofofthe
themost
most representative
representative PHD2/HIFs-α
PHD2/HIFs-αcomplexes conformations,
complexes with
conformations, with
the β2β3-loop colored in teal, C-terminal in green, HIF-1α in orange, and HIF-2α in purple.
the β2β3-loop colored in teal, C-terminal in green, HIF-1α in orange, and HIF-2α in purple. On theOn
left, a bird’s-eye view obtained with a 90° rotation of the complexes helps visualize the binding
the left, a bird’s-eye view obtained with a 90◦ rotation of the complexes helps visualize the binding
pocket. (A) PHD2/HIF-1α. (B) PHD2/HIF-2α.
pocket. (A) PHD2/HIF-1α. (B) PHD2/HIF-2α.
2.4. Binding Free Energy Analysis
2.4. Binding Free Energy Analysis
Our results showed that in the PHD2/HIF-2α complex, the C-terminus tends to
Our results showed that in the PHD2/HIF-2α complex, the C-terminus tends to
“close” the binding pocket by forming an interaction with the substrate and the β2β3-
“close” the binding pocket by forming an interaction with the substrate and the β2β3-
loop. This behavior, which is not observed with HIF-1α, could indicate a difference in
loop. This behavior, which is not observed with HIF-1α, could indicate a difference in
PHD2 substrate specificity. To deepen this observation, we performed binding-free
PHD2 substrate specificity. To deepen this observation, we performed binding-free energy
energy calculations on the representative conformations obtained from the six most
calculations on the representative conformations obtained from the six most populated
populated clusters (Table S1). Systems with HIF-1α present higher values of negative
clusters (Table S1). Systems with HIF-1α present higher values of negative binding energy
binding energy than those with HIF-2α, regardless of TPO, thus indicating a greater
than
affinity ofwith
those PHD2 HIF-2α, regardless
for HIF-1α, as alsoofreported
TPO, thus indicating
in the a Interestingly,
literature. greater affinity weofobserved
PHD2 for
HIF-1α, as also increase
a remarkable reportedininthe
the
ΔGliterature.
value forInterestingly, we observed
those conformations showing a remarkable
a C-terminusincrease
that
the ∆G
incloses the binding pocket by interacting with either HIF-2α or the ꞵ2ꞵ3-loop. binding
value for those conformations showing a C-terminus that closes the These
pocket by suggest
findings interacting with interactions
that these either HIF-2α or the β2β3-loop.
destabilize the complexThese findings
differently fromsuggest that
our initial
these interactions destabilize the complex differently from our initial hypothesis.
hypothesis. Similarly, we found that Thr405 phosphorylation causes a variation in the Similarly,
webinding
found thatfreeThr405
energyphosphorylation
despite havingcausesa limited structural
a variation in theimpact
bindingon freePHD2/HIFs-α
energy despite
complexes. To discriminate whether the observed differences in ΔG were significant,
having a limited structural impact on PHD2/HIFs-α complexes. To discriminate we
whether
theperformed
observedadifferences ∆G
t-test. Figure 8 shows the p-values obtained by comparing all the complexes.
in were significant, we performed a t-test. Figure 8 shows
the p-values obtained by comparing all the complexes. In particular, it was observed that
there is no statistically significant difference (p > 0.05) between the average binding energy
among replicas of the same complex (Figure 8A,B). This finding assumes a particularly
important value for the PHD2/HIF-2α complex, as it suggests that despite being a model
and presumably having inherent variability, its dynamic behavior remains consistent across
all simulations. A statistically significant difference (p-value < 0.05) was found comparing
PHD2/HIF-1α and PHD2/HIF-2α complexes (Figure 8C), indicating a different substrate
specificity. A p-value < 0.05 was also found comparing the phosphorylated PHD2/HIF-1α
Int. J. Mol. Sci. 2023, 24, 4710 11 of 15

and PHD2/HIF-2α complexes (Figure 8D). However, between the same phosphorylated
and non-phosphorylated complex, a p-value > 0.05 was found (Figure 8E,F). We speculate
Int. J. Mol. Sci. 2023, 24, x FOR PEER REVIEW
that the significant difference observed among the phosphorylated complexes13 ofis18due to

PHD2 substrate specificity and not induced by the presence of the phosphorylation itself.

Figure 8. Scatter dot plots and bar plots (mean with SEM) representing the difference in the mean
value8.ofScatter
Figure dot plots
the binding and bar
free energy plots
(ΔG) (mean
among with
all the SEM) representing
complexes. the plot
(A) A scatter dot difference in the mean
for ΔG value
among
value of thethebinding
four replicas of the PHD2/HIF-1α
free energy (∆G) among all complex, N = 24; 1st
the complexes. MD
(A) A (−115.1
scatter ±dot
5.356);
plot2nd
forMD∆G value
(−115.1
among the± four
2.603); 3rd MDof(−116.5
replicas ± 4.979); and 4thcomplex,
the PHD2/HIF-1α MD (−99.15
N =± 24;
9.370).
1st (B)
MD Scatter dot ±
(−115.1 plot for ΔG
5.356); 2nd MD
Int. J. Mol. Sci. 2023, 24, 4710 12 of 15

(−115.1 ± 2.603); 3rd MD (−116.5 ± 4.979); and 4th MD (−99.15 ± 9.370). (B) Scatter dot plot for
∆G value among the three replicas of the PHD2/HIF-2α complex, N = 18; 1st MD (−90.64 ± 7.545);
2nd MD (−100.1 ± 5.448); and 3rd MD (−80.55 ± 7.821). (C) A scatter dot plot of the ∆G value shows
−111.5 ± 3.204, N = 24 in PHD2/HIF-1α and −90.43 ± 4.270, N = 18 in PHD2/HIF-2α complexes.
*** = p-value ≤ 0.001 (D) A scatter dot plot for the ∆G value shows −117.7 ± 7.250, N = 6 in PHD2/HIF-
1α (TPO) and −85.25 ± 2.537, N = 6 in PHD2/HIF-2α (TPO) complexes. ** = p-value ≤ 0.01 (E) A bar
plot for ∆G value −111.5 ± 3.204, N = 24 in PHD2/HIF-1α and −117.7 ± 7.250, N = 6 in PHD2/HIF-
1α (TPO) complexes. (F) A bar plot for ∆G values −90.43 ± 4.270, N = 18 in PHD2/HIF-2α and
−85.25 ± 2.537, N = 6 in PHD2/HIF-2α (TPO) complexes.

3. Discussion
In this work, we investigated the substrate specificity of the PHD2 enzyme in complex
with the transcription factors HIF-1α and HIF-2α. PHD2 is a well-known trigger of the
adaptive hypoxic response, and its enzymatic deregulation is linked to multiple human
diseases, such as polycythemia and cancer [27]. This enzyme presents a different substrate
specificity; however, the molecular details of this behavior are still poorly understood.
Our investigations showed that residue-residue interactions between PHD2/HIF-1α and
PHD2/HIF-2α are mostly conserved; however, they also suggest that the PHD2 C-terminus
may play a role in favoring the interaction with specific substrates. Further reinforcing
this proposed role, we report a direct interaction of the PHD2 C-terminus with HIF-2α
that is not observed when the protein is in complex with HIF-1α. We also inspected the
possible effect induced by the phosphorylation of Thr405, described in leukemia cells
and localizing on the PHD2 C-terminal tail. Molecular dynamics simulations showed
no significant difference in the stabilities of complexes upon phosphorylation; rather, we
observed that the phosphorylated C-terminus engages in direct interaction with the PHD2
β2β3-loop. Although this interaction is not stably maintained for the entire simulation,
it indicates a C-terminus tendency to close the binding pocket by interacting with the
β2β3-loop and possibly acting as a molecular switch to activate/inactivate the enzyme.
A similar tendency to close the binding pocket was also registered for the PHD2/HIF-2α
complex. Binding energy calculations indicate that this closure of the active site by the
PHD2 C-terminus increases the energy value, indicating that this specific interaction may
reduce the substrate affinity by destabilizing the complex. We also observed a significant
difference in the binding energy between the PHD2/HIF-1α and PHD2/HIF-2α complexes.
Although a certain degree of variability is linked with the final conformation assumed by
the complexes during simulations, particularly the C-terminus tendency to form an extra
α-helix, we believe that the differences in binding energy between the two substrates may
be due to an important entropic effect in the complexes formed by HIF-2α.

4. Materials and Methods

4.1. Homology Modeling
The homology model of HIF-2α was performed using Modeller [28], since its X-ray
structure exists in the hydroxylated form in complexes with the pVHL-Elongin C–Elongin
B (VCB) complex (PDB ID: 6I7R). Therefore, we used the crystal structure of HIF-1α CODD
(556–574) in complex with PHD2 (PDB ID: 6YW3) as a template. T-Coffee [29] was used to
align the sequences. Ten models were generated, and the best-scoring one was selected
using the DOPE score and subsequently modeled on HIF-1α into the PHD2 binding pocket.
In order to avoid steric clashes among residues, a minimization run followed by 2 ns of
NVT simulations/ensembles were performed.

4.2. Molecular Dynamics Simulations

The crystal structure of PHD2 in closed conformation (PDB ID: 6YW3) was used
as the starting structure. For all systems simulated, N-oxalylglycine (NOG) inhibitor
was replaced with α-ketoglutarate (AKG), the natural co-substrate, by superimposing the
crystal structure of PHD2 (PDB ID: 6YW1) in complex with it. The manganese (II) ion
Int. J. Mol. Sci. 2023, 24, 4710 13 of 15

(Mn+2 ) in the native structure was modified to ferrous ion (Fe+2 ). All MD simulations
were performed with GROMACS (2020.6) [30], using the CHARMM36 force field [31]. The
AKG parameters are not included in the force field, so we generated the corresponding
parameters with CHARMM-GUI [32] and CGenFF [33], respectively, and implemented
them in the CHARMM36m force field. As AKG contains two carboxyl groups that are
deprotonated at pH 7, we recalculated the partial atomic charges using those of the amino
acids glutamate and glutamine inserted in the force field as references and charges obtained
with MOPAC [34] (Figure S1). A cubic box with a distance of 10 Å was generated, filled with
TIP3P water molecules, and ionized with 0.15 M NaCl. The system was minimized using
a steepest descent algorithm followed by 2 ns of NVT ensemble, 2 ns of NPT ensemble,
and then by 1 µs of classical molecular dynamics (MD) simulation. The temperature was
coupled with a V-rescale thermostat and maintained at 300 K, while in the NPT simulation,
a Berendsen barostat was used. The pressure was maintained at 1 atm. Three independent
replicas were obtained for all the systems, while a fourth replica was specifically calculated
for the PHD2/HIF-1α complex. This further simulation and the following statistical
analysis were considered necessary to strengthen our results. Indeed, the first simulation
of this specific system presented some outliers, likely due to this trajectory being obtained
from an extension up to 1 µs of an initial 500 ns run. The RMSD, RMSF, and radius of
gyration were calculated with GROMACS, and plots were generated using Grace [35].

4.3. Phosphorylation of PHD2-Thr405

Information on the phosphorylation of the Threonine405 residue located on the PHD2
C-terminus was retrieved by PhosphoSitePlus [26]. This post-translational modification
(PTM) was characterized in vivo by mass spectrometry. The phosphate group was added
to the Threonine405 residue using UCSF Chimera’s Build structure tool [36]. A 200 ns
molecular dynamics simulation was then performed to further stabilize the system. The
last frame of the resulting simulation was used as the starting conformation to run a 1 µs
molecular dynamics simulation.

4.4. Cluster Analysis

The cluster analysis was performed for all simulations using the RING PyMol plu-
gin [37], on a local Linux workstation. Clusters were generated on Cα by imposing a
RMSD threshold of 3 Å. For each system, the first 6 most representative clusters were
selected to obtain an equal and comparable number. Subsequently, the most representative
conformation of each cluster was used to calculate the binding free energy (see Section 4.5).

4.5. Binding Free Energy and the T-Test

Binding free energies were calculated using MM/GBSA free energy decomposition
methods integrated into HawkDock [38], a web server that predicts and analyzes protein-
protein interactions. The most representative conformation of each cluster was used to
carry out the analysis. Calculations were repeated for all 6 conformations derived from
each simulation, and the average of the resulting binding energy values was calculated
(Table S1). A t-test was used to test whether the means of two populations were significantly
different. In particular, a two-sample t-test was performed for groups with homogeneous
variances and a Welch’s t-test or an unequal variances t-test for those with unequal variances
(Tables S3 and S4). A Shapiro–Wilk test preceded this analysis to verify the assumption
that each sample size was normally distributed (Table S2).

5. Conclusions
Our data, although generated in silico, confirm a greater affinity for HIF-1α than
HIF-2α for PHD2. They also suggest that the PHD2 C-terminus could act as a molecular
regulator of the enzyme activity. Furthermore, this investigation highlights specific residues
that allow PHD2 to discriminate between HIF-1α and HIF-2α. Considering the PHDs’ role
in human diseases, the identification of these sites may be of relevance for cancer research
Int. J. Mol. Sci. 2023, 24, 4710 14 of 15

as their mutations can interfere with the binding of a specific substrate without impacting
the PHD2 enzymatic activity.

Supplementary Materials: The following supporting information can be downloaded at: https:
//www.mdpi.com/article/10.3390/ijms24054710/s1.
Author Contributions: Conceptualization, G.M. and S.C.E.T.; methodology, G.F.C. and G.M.; inves-
tigation, G.F.C.; formal analysis and validation, G.F.C. and G.M.; data curation and bioinformatics
analysis, G.F.C. and G.M.; writing—original draft preparation and editing, G.F.C., G.M. and S.C.E.T.
All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by Fondazione AIRC per la Ricerca sul Cancro (AIRC), grant
number IG 2019 ID 23825.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The datasets generated during and/or analyzed during the current
study are available from the corresponding author on reasonable request.
Acknowledgments: We acknowledge Franco Pradelli for the helpful discussions.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or
in the decision to publish the results.

References
1. Semenza, G.L. Oxygen Sensing, Homeostasis, and Disease. N. Engl. J. Med. 2011, 365, 537–547. [CrossRef] [PubMed]
2. Yang, M.; Su, H.; Soga, T.; Kranc, K.R.; Pollard, P.J. Prolyl Hydroxylase Domain Enzymes: Important Regulators of Cancer
Metabolism. Hypoxia 2014, 2, 127–142. [CrossRef] [PubMed]
3. Epstein, A.C.R.; Gleadle, J.M.; McNeill, L.A.; Hewitson, K.S.; O’Rourke, J.; Mole, D.R.; Mukherji, M.; Metzen, E.; Wilson, M.I.;
Dhanda, A.; et al. C. Elegans EGL-9 and Mammalian Homologs Define a Family of Dioxygenases That Regulate HIF by Prolyl
Hydroxylation. Cell 2001, 107, 43–54. [CrossRef] [PubMed]
4. Masson, N.; Willam, C.; Maxwell, P.H.; Pugh, C.W.; Ratcliffe, P.J. Independent Function of Two Destruction Domains in
Hypoxia-Inducible Factor-Alpha Chains Activated by Prolyl Hydroxylation. EMBO J. 2001, 20, 5197–5206. [CrossRef]
5. Ivan, M.; Kondo, K.; Yang, H.; Kim, W.; Valiando, J.; Ohh, M.; Salic, A.; Asara, J.M.; Lane, W.S.; Kaelin, W.G. HIFalpha Targeted
for VHL-Mediated Destruction by Proline Hydroxylation: Implications for O2 Sensing. Science 2001, 292, 464–468. [CrossRef]
6. Prabhakar, N.R.; Semenza, G.L. Adaptive and Maladaptive Cardiorespiratory Responses to Continuous and Intermittent Hypoxia
Mediated by Hypoxia-Inducible Factors 1 and 2. Physiol. Rev. 2012, 92, 967–1003. [CrossRef]
7. Feldman, D.E.; Thulasiraman, V.; Ferreyra, R.G.; Frydman, J. Formation of the VHL–Elongin BC Tumor Suppressor Complex Is
Mediated by the Chaperonin TRiC. Mol. Cell 1999, 4, 1051–1061. [CrossRef]
8. Melville, M.W.; McClellan, A.J.; Meyer, A.S.; Darveau, A.; Frydman, J. The Hsp70 and TRiC/CCT Chaperone Systems Cooperate
in Vivo to Assemble the von Hippel-Lindau Tumor Suppressor Complex. Mol. Cell. Biol. 2003, 23, 3141–3151. [CrossRef]
9. Kamura, T.; Sato, S.; Iwai, K.; Czyzyk-Krzeska, M.; Conaway, R.C.; Conaway, J.W. Activation of HIF1α Ubiquitination by
a Reconstituted von Hippel-Lindau (VHL) Tumor Suppressor Complex. Proc. Natl. Acad. Sci. USA 2000, 97, 10430–10435.
[CrossRef]
10. Ohh, M.; Park, C.W.; Ivan, M.; Hoffman, M.A.; Kim, T.Y.; Huang, L.E.; Pavletich, N.; Chau, V.; Kaelin, W.G. Ubiquitination of
Hypoxia-Inducible Factor Requires Direct Binding to the Beta-Domain of the von Hippel-Lindau Protein. Nat. Cell Biol. 2000, 2,
423–427. [CrossRef]
11. Minervini, G.; Mazzotta, G.M.; Masiero, A.; Sartori, E.; Corrà, S.; Potenza, E.; Costa, R.; Tosatto, S.C.E. Isoform-Specific Interactions
of the von Hippel-Lindau Tumor Suppressor Protein. Sci. Rep. 2015, 5, 12605. [CrossRef] [PubMed]
12. Iliopoulos, O.; Kibel, A.; Gray, S.; Kaelin, W.G. Tumour Suppression by the Human von Hippel-Lindau Gene Product. Nat. Med.
1995, 1, 822–826. [CrossRef] [PubMed]
13. Barry, R.E.; Krek, W. The von Hippel-Lindau Tumour Suppressor: A Multi-Faceted Inhibitor of Tumourigenesis. Trends Mol. Med.
2004, 10, 466–472. [CrossRef] [PubMed]
14. Gossage, L.; Eisen, T.; Maher, E.R. VHL, the Story of a Tumour Suppressor Gene. Nat. Rev. Cancer 2015, 15, 55–64. [CrossRef]
[PubMed]
15. Ohh, M.; Taber, C.C.; Ferens, F.G.; Tarade, D. Hypoxia-Inducible Factor Underlies von Hippel-Lindau Disease Stigmata. eLife
2022, 11, e80774. [CrossRef] [PubMed]
16. Richardson, N.L.; O’Malley, L.J.; Weissberger, D.; Tumber, A.; Schofield, C.J.; Griffith, R.; Jones, N.M.; Hunter, L. Discovery of
Neuroprotective Agents That Inhibit Human Prolyl Hydroxylase PHD2. Bioorganic Med. Chem. 2021, 38, 116115. [CrossRef]
Int. J. Mol. Sci. 2023, 24, 4710 15 of 15

17. Hypoxia-Inducible Factor 1 Is a Basic-Helix-Loop-Helix-PAS Heterodimer Regulated by Cellular O2 Tension. Available online:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC41725/ (accessed on 18 November 2022).
18. Berra, E.; Benizri, E.; Ginouvès, A.; Volmat, V.; Roux, D.; Pouysségur, J. HIF Prolyl-Hydroxylase 2 Is the Key Oxygen Sensor
Setting Low Steady-State Levels of HIF-1alpha in Normoxia. EMBO J. 2003, 22, 4082–4090. [CrossRef]
19. Pan, Y.; Mansfield, K.D.; Bertozzi, C.C.; Rudenko, V.; Chan, D.A.; Giaccia, A.J.; Simon, M.C. Multiple Factors Affecting Cellular
Redox Status and Energy Metabolism Modulate Hypoxia-Inducible Factor Prolyl Hydroxylase Activity in Vivo and in Vitro. Mol.
Cell Biol. 2007, 27, 912–925. [CrossRef]
20. Minervini, G.; Quaglia, F.; Tosatto, S.C.E. Insights into the Proline Hydroxylase (PHD) Family, Molecular Evolution and Its Impact
on Human Health. Biochimie 2015, 116, 114–124. [CrossRef]
21. Strowitzki, M.J.; Cummins, E.P.; Taylor, C.T. Protein Hydroxylation by Hypoxia-Inducible Factor (HIF) Hydroxylases: Unique or
Ubiquitous? Cells 2019, 8, 384. [CrossRef]
22. Nguyen, T.L.; Durán, R.V. Prolyl Hydroxylase Domain Enzymes and Their Role in Cell Signaling and Cancer Metabolism. Int J.
Biochem. Cell Biol. 2016, 80, 71–80. [CrossRef] [PubMed]
23. Meneses, A.M.; Wielockx, B. PHD2: From Hypoxia Regulation to Disease Progression. Hypoxia 2016, 4, 53–67. [CrossRef]
[PubMed]
24. Gaete, D.; Rodriguez, D.; Watts, D.; Sormendi, S.; Chavakis, T.; Wielockx, B. HIF-Prolyl Hydroxylase Domain Proteins (PHDs) in
Cancer-Potential Targets for Anti-Tumor Therapy? Cancers 2021, 13, 988. [CrossRef] [PubMed]
25. Yu, M.; Lun, J.; Zhang, H.; Zhu, L.; Zhang, G.; Fang, J. The Non-Canonical Functions of HIF Prolyl Hydroxylases and Their Dual
Roles in Cancer. Int. J. Biochem. Cell Biol. 2021, 135, 105982. [CrossRef] [PubMed]
26. Hornbeck, P.V.; Zhang, B.; Murray, B.; Kornhauser, J.M.; Latham, V.; Skrzypek, E. PhosphoSitePlus, 2014: Mutations, PTMs and
Recalibrations. Nucleic Acids Res. 2015, 43, D512–D520. [CrossRef] [PubMed]
27. Strocchi, S.; Reggiani, F.; Gobbi, G.; Ciarrocchi, A.; Sancisi, V. The Multifaceted Role of EGLN Family Prolyl Hydroxylases in
Cancer: Going beyond HIF Regulation. Oncogene 2022, 41, 3665–3679. [CrossRef]
28. Sali, A.; Blundell, T.L. Comparative Protein Modelling by Satisfaction of Spatial Restraints. J. Mol. Biol. 1993, 234, 779–815.
[CrossRef]
29. Notredame, C.; Higgins, D.G.; Heringa, J. T-Coffee: A Novel Method for Fast and Accurate Multiple Sequence Alignment. J. Mol.
Biol. 2000, 302, 205–217. [CrossRef]
30. Bekker, H.; Berendsen, H.; Dijkstra, E.; Achterop, S.; Vondrumen, R.; Vanderspoel, D.; Sijbers, A.; Keegstra, H.; Renardus, M.
GROMACS—A parallel computer for molecular-dynamics simulations. In Proceedings of the 4th International Conference on
Computational Physics (PC 92), Prague, Czech Republic, 24–28 August 1992; pp. 252–256.
31. Vanommeslaeghe, K.; Hatcher, E.; Acharya, C.; Kundu, S.; Zhong, S.; Shim, J.; Darian, E.; Guvench, O.; Lopes, P.; Vorobyov, I.;
et al. CHARMM General Force Field: A Force Field for Drug-like Molecules Compatible with the CHARMM All-Atom Additive
Biological Force Fields. J. Comput. Chem. 2010, 31, 671–690. [CrossRef]
32. Jo, S.; Kim, T.; Iyer, V.G.; Im, W. CHARMM-GUI: A Web-based Graphical User Interface for CHARMM. J. Comput. Chem. 2008, 29,
1859–1865. [CrossRef]
33. Vanommeslaeghe, K.; MacKerell, A.D., Jr. Automation of the CHARMM General Force Field (CGenFF) I: Bond Perception and
Atom Typing. J. Chem. Inf. Model. 2012, 52, 3144–3154. [CrossRef] [PubMed]
34. Stewart, J.J.P. MOPAC2016; Stewart Computational Chemistry: Colorado Springs, CO, USA, 2016; Available online: http:
//OpenMOPAC.Net (accessed on 20 January 2022).
35. Turner, P.J.; Turner, P.J. XMGRACE, Version 5.1. 19; Center for Coastal and Land-Margin Research, Oregon Graduate Institute of
Science and Technology: Beaverton, OR, USA, 2005.
36. Pettersen, E.F.; Goddard, T.D.; Huang, C.C.; Couch, G.S.; Greenblatt, D.M.; Meng, E.C.; Ferrin, T.E. UCSF Chimera–a Visualization
System for Exploratory Research and Analysis. J. Comput. Chem. 2004, 25, 1605–1612. [CrossRef] [PubMed]
37. Clementel, D.; Del Conte, A.; Monzon, A.M.; Camagni, G.F.; Minervini, G.; Piovesan, D.; Tosatto, S.C.E. RING 3.0: Fast generation
of probabilistic residue interaction networks from structural ensembles. Nucleic Acids Res. 2022, 50, W651–W656. [CrossRef]
[PubMed]
38. HawkDock: A Web Server to Predict and Analyze the Protein–Protein Complex Based on Computational Docking and
MM/GBSA|Nucleic Acids Research|Oxford Academic. Available online: https://academic.oup.com/nar/article/47/W1
/W322/5491747 (accessed on 18 November 2022).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.