Добавил:
Sekretar
kiopkiopkiop18@yandex.ru
t.me/Prokururor I Вовсе не секретарь, но почту проверяю
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз:
Предмет:
Файл:Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5319_Библиотеки_им_академика_М_И_Перельмана.pdf
X
- •Preface and Acknowledgement
- •Chemical Structures of Amino Acids,Molecular Graphics and Introduction
- •Introduction
- •Literature
- •Chapter Abstract Videos
- •Contents
- •About the author
- •1.10 Synopsis
- •1.3 The Battle Against Infectious Disease
- •1.4 Biological Concepts in Drug Research
- •Bibliography and Further Reading
- •2.8 A Long List of Accidents
- •2.10 Synopsis
- •Bibliography and Further Reading
- •3. Classical Drug Research
- •3.2 Malaria: Success and Failure
- •3.6 Synopsis
- •Bibliography and Further Reading
- •4.1 The Lock-and-Key Principle
- •4.2 The Essential Role of the Membrane
- •4.6 Blame It All on Water!
- •4.11 Lessons for Drug Design
- •4.12 Synopsis
- •Bibliography and Further Reading
- •5.1 Louis Pasteur Sorts Crystals
- •5.2 Structural Basis of Optical Activity
- •5.4 Lipases Separate Racemates
- •5.8 Synopsis
- •Bibliography and Further Reading
- •6.2 Lead Structures from Plants
- •6.9 Synopsis
- •Bibliography and Further Reading
- •7.2 Color Change Demonstrates Activity
- •7.7 Biophysics Supports Screening
- •7.11 Synopsis
- •Bibliography and Further Reading
- •8.1 Strategies for Drug Optimization
- •8.5 From Agonists to Antagonists
- •8.9 Synopsis
- •Bibliography and Further Reading
- •9. Designing Prodrugs
- •9.1 Foundations of Drug Metabolism
- •9.2 Esters Are Ideal Prodrugs
- •9.6 Synopsis
- •Bibliography and Further Reading
- •10. Peptidomimetics
- •10.1 Therapeutic Relevance of Peptides
- •10.2 Designing Peptidomimetics
- •Bibliography and Further Reading
- •11.4 What Is Contained in Chemical Space?
- •Bibliography and Further Reading
- •12.7 Silencing Genes by RNA Interference
- •12.9 Proteomics and Metabolomics
- •Bibliography and Further Reading
- •13.3 Crystal Lattices Diffract X-Rays
- •Bibliography and Further Reading
- •Bibliography and further reading
- •15. Molecular Modeling
- •15.2 Strategies in Molecular Modeling
- •15.3 Knowledge-Based Approaches
- •15.4 Force Field Methods
- •15.5 Quantum Chemical Methods
- •Bibliography and further reading
- •16. Conformational Analysis
- •16.8 Synopsis
- •Bibliography and Further Reading
- •Bibliography and Further Reading
- •18.4 Lipophilicity and Biological Activity
- •Bibliography and Further Reading
- •19.3 The Role of Hydrogen Bonds
- •19.5 Absorption Profiles of Acids and Bases
- •19.8 From In Vitro to In Vivo Activity
- •Bibliography and Further Reading
- •Bibliography and Further Reading
- •21.5 LUDI Discovers the First Leads
- •Bibliography and Original Papers
- •22.1 The Druggable Genome
- •22.4 Enzymes and Their Inhibitors
- •22.9 Resistance and Its Origin
- •Bibliography and Further Reading
- •23.1 Serine-Dependent Hydrolases
- •23.10 Synopsis
- •Bibliography and Further Reading
- •24. Aspartic Protease Inhibitors
- •24.2 Design of Renin Inhibitors
- •24.8 Synopsis
- •Bibliography and Further Reading
- •25.1 Structure of Zinc Metalloproteases
- •25.9 What Zinc Can Do, Iron Can Too
- •25.11 Synopsis
- •Bibliography and Further Reading
- •26. Transferase Inhibitors
- •26.1 The Kinase “Gold Rush”
- •Bibliography and Further Reading
- •27. Oxidoreductase Inhibitors

. • Tracking the Dynamic Transformation with the Appropriate Spins
21.13 Only Serendipity Can Help:
Different Crystal Form—
New Dimer
While searching for new inhibitors addressing the preQ1
binding site with sugar-like side chains, Frederik Ehrmann made asurprising observation: from the same
crystallization sample, the enzyme with these inhibitors
crystallized side by side in two different crystal forms!
On the one hand, we found the well-known crystal form
in space group C2. It showed the usual functional dimer
with the dimer packing described above. In the other
crystal form, now in space group P21, the dimer adopts
adifferent mutual interface packing. Here, the structurally almost unchanged monomer units form anew contact surface interface in an orientation rotated by about
130° with respect to the other dimer form. We will call
this form the “twisted” homodimer (. Fig.21.21, right).
Remarkably, the size and number of interface contacts
were hardly reduced. However, what is critical to the disruption of enzyme function is that tRNA binding is no
longer possible to the twisted dimer for steric reasons.
Thus, this new dimer packing, apparently induced by
the binding of our novel inhibitors, freezes the protein
in an inactive state. As aresult, it is no longer able to
induce catalytic turnover. The inhibitor 21.14 had been
found previously, but had only been studied structurally
by crystal soaking. This method, which works with premanufactured crystals, naturally yielded the usual C2
homodimer. Now, following acocrystallization protocol
where the ligand is added already to the crystallization
solution, the twisted homodimer was found in the crystals in space group P21. Thus, we had to learn that the
applied crystallization protocol used initially led to incomplete conclusions.
Since both crystal forms grow side by side from the
same crystallization well, we assume that both forms
coexist in solution and have very similar stability. However, crystallography only determines the end points of
this assumed rotational transformation between the two
forms. Therefore, the intriguing questions were: does this
transformation also occur in solution, does it depend on
the ligand used, and are there structural features that indicate what causes the transformation? If there is asubstance that stabilizes the twisted, catalytically inactive
form, it will also be acandidate for drug development as
it will also block the biological function of TGT. Such
asubstance will indeed be able to keep our enzyme in
acatalytically inactive state. Closely related to this is the
question of why the enzyme is able to adopt such astate
in the rst place. Is this astate of self-regulation and
is the enzyme “slowed down” in its catalytic activity by
certain ligands that may be undesirable or present in too
high aconcentration?
21.14 Tracking the Dynamic
Transformation with
the Appropriate Spins
In solution, methods that observe the spins of magnetic nuclei and their coupling to each other are very
powerful techniques for structure determination (see
Sects.7.8 and13.7). While NMR spectroscopy detects
spin couplings only over relatively short distances,
the spins of unpaired electrons can couple over much
longer distances. Thus, electron paramagnetic spin
resonance (EPR) spectroscopy seemed ideally suited
to study the behavior of TGT in solution. EPR spec-
troscopy requires unpaired electrons in the molecule.
Since paramagnetic centers are absent in most biomolecules, spin labels must be introduced into the biological
system in asite-specic manner. If two or more spin
labels are present in amolecule, so-called pulsed electron–electron double resonance (PELDOR or DEER)
experiments can be used to obtain precise information
about changes in the distance between the spin labels in
the range of 15–60 Å. This is exactly the range in which
we expected the changes to occur during the transformation of the TGT between the two dimeric states. Ni-
troxide spin labels are most commonly used for this purpose. They can be selectively coupled via the thiol group
of acysteine residue in the form of adisulde bridge
(. Fig.21.22). Site-specicity on the protein is achieved
by mutating cysteine residues into the protein sequence
at the desired positions. At the same time, undesired
cysteines found in the wild-type sequence are replaced
by, for example, alanine or serine. The search for the
best spin-labeling sites on the TGT dimer in both of its
forms was initially planned with the help of acomputer
simulation.
Dzung Nguyen from the group in Marburg and Dinar Abdullin from Olav Schiemann’s group in Bonn,
Germany, found positions87 and 319 to be particularly
suitable for spin labels. After the protein variants had
been prepared and labeled, and the equilibria with the
different ligands had been established in solution, the
samples were shock-frozen at −196 °C and measured.
This gave the result shown in . Fig.21.23.
Obviously, ligand21.27 predominantly induces the
twisted form of the enzyme. It can, therefore, be considered as astabilizer of the inactive twisted form. Such
acompound is expected to inhibit the enzyme by acompletely different mechanism. Time-dependent measurements were also performed for this ligand using the PELDOR method. After only one hour of equilibration in
solution, asignicant amount (> 70%) of twisted dimer
is present. After 24 h, the nal equilibrium is reached
with more than 85% of this form.
EPR spectroscopy is not the only way to study solution equilibrium. 19F-NMR spectroscopy is also avery
sensitive method for tracking structural differences in

Chapter • A Case Study: Structure-Based Inhibitor Design for tRNA-Guanine Transglycosylase
. Fig. 21.22 A nitroxide spin label is attached via adisulde bridge
to the thiol group of an appropriately positioned cysteine residue. In
the homodimer, the distance between the two labels can now be measured in the range of 15–60 Å. At positions87 and 319, aglycine and
ahistidine, respectively, are exchanged for cysteines. In the functional
(yellow arrow) or twisted (green arrow) dimer, the distance between the
labels changes from about 55 to 25 and 30 to 60 Å, respectively. Because of the conformational exibility of the nitroxide group at the
protein surface, adistribution in anarrow distance range is detected
(dark blue scattered distribution)
21
. Fig. 21.23 Left For position87, calculations based on the crystal
structures of the functional and twisted homodimer give a distribution of about 55 and 25 Å with 21.14 and 21.27, respectively. The experiment shows that TGT is present in the functional form without
any bound ligand or with bound tRNA. Ligand21.21 without aside
chain also binds only to TGT in the functional dimer form. For ligands 21.29, 21.14, and especially 21.27, the relative proportion of the
solution. Fluorine atoms are generally hardly present
in biological structures. Therefore, they have to be incorporated articially via uorine-labeled amino acids.
We have chosen auorinated tryptophan substituted at
position5. TGT contains four tryptophan residues per
monomer unit. Of these, Trp95 and Trp 326 are located
near the dimer interface. The other two residues, Trp
178 and Trp 296, are more distant. Trp 178 is completely
buried inside the protein, while Trp 296 is partially oriented towards the surface. To introduce the uorinated
twisted dimer population increases. For 21.27, up to 85% is present in
the twisted form. Right Position 319 conrms this nding. Here the
calculated distance in the functional dimer is about 30 Å and increases
to 60 Å in the twisted form. Experimentally, there is conrmation that
21.27 is astabilizer of the twisted form. The distribution in this case is
more complex and shows two maxima, suggesting two different conformational families for the attached spin labels
residues, Andreas Nguyen used the expression of atryptophan-auxotrophic cell line that cannot produce the aromatic amino acid itself. After adding the uorinated
amino acid in place of the natural one to the expression
medium and starting protein expression, the 19F-labeled
amino acid is incorporated into the target protein by
the cellular machinery. The correct incorporation was
subsequently veried by both crystallography and mass
spectrometry.

. • Tracking the Dynamic Transformation with the Appropriate Spins
. Fig. 21.24 Upper left 19F-NMR spectra of TGT
different Trp ⇨ Phe mutants of the wild type, Trp95Phe, Trp178Phe,
and Trp326Phe variants. Two low molecular weight contaminants (uoride ions and free 5F-Trp) are indicated by an asterisk. The absence
of individual signals in the Phe variants assigns the resonances to the
individual Trp residues. Bottom left 19F-NMR spectra of TGT
without (TGT
21.14–21.32. For 5F-Trp178, 5F-Trp296, and 5F-Trp326 with ligands
21.20 and 21.30 (green background) hardly or only slightly changed
, lowest spectrum) and with the listed ligands
5F-Trp
5F-Trp
and three
5F-Trp
We performed the NMR studies together with Michael Sattler’s group at the TU Munich, Germany. As
expected, the spectra indicated four different uorine resonances. First, the different 19F-labeled tryptophans had
to be assigned to the signals. This was done by successively replacing the tryptophans with phenylalanine and
observing which signals disappeared from the spectrum
(. Fig.21.24, top).
Subsequently, different ligands from the series
21.14–21.32 were titrated into asolution of the labeled
TGT
until maximally a three-fold molar excess
5F-Trp
of the ligands was present. Depending on the ligand
used, different amounts of changes in the resonance
resonance positions are observed compared to the uncomplexed protein. For both, no evidence for the formation of the twisted dimer
could be detected crystallographically. For ligands 21.14, 21.27, and
21.28, however, structures of both forms could be crystallized. With
these ligands similar spectra are observed, which are strongly altered
compared to the uncomplexed protein (red background). But also the
spectra with the ligands 21.31, 21.32, and 21.29 show clear changes
compared to the uncomplexed protein. Here, both forms are probably
present in solution
positions were observed. Using the unliganded protein
as areference (. Fig.21.24, bottom spectrum), which
according to all previous results is present in solution as
the functional C2 symmetric homodimer, its spectrum
shows similarities to that of the complex with 21.20.
Only Trp 95 undergoes a signicant shift. The other
residues, especially Trp 326 in the interface contact region, remain virtually unchanged. The situation is very
similar for ligand21.30. Ligand21.20 lacks aside chain
and there is no evidence that this molecule triggers the
transformation to the inactive, twisted form. Similarly,
no transformation was observed for ligand21.30, despite
its close similarity to 21.27 and 21.28. According to the

Chapter • A Case Study: Structure-Based Inhibitor Design for tRNA-Guanine Transglycosylase
21
. Fig. 21.25 Left Binding mode of ligand21.27, which stabilizes the
twisted form of TGT. In the ligand-free state, the homodimer adopts
the geometry of the functional dimer. When the enzyme adopts the
structure of the twisted dimer, the helix αA shifts from its position
in the ligand-free state (gray) to the geometry of the twisted dimer
(ochre). In the structure with 21.27, corresponding to the geometry
of the functional dimer, this helix is completely disordered (yellow
dashed line). When compared to the geometry of the uncomplexed
TGT (gray), we see that the anomeric methoxy group of 21.27 push-
PELDOR measurements, ligand21.27 is avery strong
stabilizer of the twisted form. Its NMR spectrum shows
clearly different resonance positions compared to those
with 21.20 and 21.30, especially for Trp 326. The same is
true for 21.14 and 21.28. For ligand21.29, both species
seem to coexist in solution. This is also true for 21.31
and 21.32, for which both the functional and the twisted
species can be detected in equilibrium.
The results of the 19F-NMR spectroscopy, thus, conrm the picture from the PELDOR studies. However,
they differentiate it even further. Not only is 21.27 able
to induce the twisted form effectively, but this raises the
question of why some ligands induce the twist, while
others do not. Acomparison of the crystal structures of
ligands 21.14, 21.27, and 21.29 provides further insight
at this point (. Fig.21.25).
To transform into the twisted dimer geometry, the
helix αA has to be shifted in its position (. Fig.21.25,
left). In the TGT complex with the geometry of the functional dimer and the ligand21.27 (yellow), the geometry
of this helix is already disturbed to such an extent that
its arrangement in the electron density can no longer be
observed. The anomeric methoxy group on the six-membered ring sugar comes into spatial conict with the side
chain of Gln 107 in the center of the helix αA. This
leads to steric stress that triggers the transformation to
the twisted dimer geometry and subsequently stabilizes
this geometry. For the other ligands that also trigger the
transformation, this steric stress is weaker (. Fig.21.25,
right). Thus, the design concept for potent stabilizers of
the catalytically inactive twisted dimer form is in place.
Ligands are needed that, on the one hand, exert efcient
steric pressure on Gln 107 to induce transformation. At
es against residue Gln 107 in the center of the helix. This triggers
the transition from the functional to the twisted dimer. Right In the
detailed view, this geometric conict of 21.27 with residue Gln 107
is again highlighted. The ligands 21.14 and 21.29, for which crystal
structures in the geometry of the functional as well as the twisted dimer could be determined, indicate asmaller spatial demand of the
4-substituent in this region. This explains why, according to the PELDOR measurements, only 21.27 proves to be an excellent stabilizer of
the twisted dimer arrangement
the same time, they must inuence the conformation of
the loop–helix motif. Finally, the barrier to return to the
functional dimer geometry must be sufciently high. It
is also possible that the discovered rearrangement mechanism is triggered by substrates or products of TGT to
bring the enzyme to aresting state. In any case, it is
important for drug development that acompletely new
inhibition mechanism of TGT has been discovered with
ligand21.27. The enzyme is blocked by putting it into
akind of “dormant” state.
21.15 When Sulfur Accidentally Oxidizes
and Starts aFragment Design
Project in aNew Arrangement
Despite our best efforts, we had not yet found away to
get an idea of the geometry of the contact interface surface in the monomeric state. By all accounts, the loop–
helix motif seemed to be important in controlling monomerization. The many crystal structures that had been
solved indicated that this motif had great conformational
exibility (. Fig.21.20, left). As mentioned above, cysteine residues were introduced to replace the amino
acids of the aromatic cluster. By this, we also aimed
to introduce as many potential attachment points for
small fragments into the contact surface as possible (cf.
Fragment Tethering Approach, Sect.7.10). Fortunately,
serendipity was once again our friend. Stephan Jakobi
made an exciting discovery: after some time, anew crystal form had grown in one of his crystallization batches.
It was the variant in which Tyr 330 had been replaced

. • When Sulfur Accidentally Oxidizes and Starts aFragment Design Project in aNew Arrangement
. Fig. 21.26 Upper row Structure of the functionally active homo-
dimer of TGT. In the aromatic cluster, Tyr 330 has been replaced by
acysteine. However, there is still ashort contact with Phe92′ of the
opposite monomeric unit (see detailed view, upper right). Bottom row
Oxygen from the air oxidizes the thiol function of the exposed Cys
330. Adisulde bridge is formed between Cys 330–Cys 330′. The enzyme is now “pseudomonomerized.” The monomer units, which have
hardly changed structurally, now pack against each other with altered
geometry (detailed view, bottom right). The former contact surface is
exposed. As aresult, the exible loop–helix motif is no longer embedded in the interface contact surface (green box). It takes anew course
by acysteine. The new crystals had ahexagonal shape
and not the usual monoclinic crystal habit! Structure
determination solved the puzzle. The original homodimer was completely altered, although the geometry of
the individual monomer units had hardly changed. Formally, the enzyme was no longer adimer. Atmospheric
oxygen had oxidized the sulfur at Cys 330, and as aconsequence, forming acovalent disulde bridge between
the two former monomer units. This new S–S bridge
was formed between Cys 330–Cys 330′. It coincides with
atwo-fold crystallographic rotation axis in the new crystal packing. Thus, it causes one monomer subunit to tilt
towards the other (. Fig.21.26). The original contact
area of over 1600 Å2 shrinks to 537 Å2. Most likely, the
(red) that differs from the original geometry in the homodimer (yellow). With this geometry, the loop no longer ts into the packing of
the former homodimer for steric reasons and, thus, blocks its formation. In order to disrupt the mutual recognition and binding of the
two monomeric units, this new conformation of the loop–helix motif
would have to be stabilized by aligand. (7 https://sn.pub/iYaYHJ)
stability of the contact area in the functional dimer of
the studied Tyr330Cys variant is signicantly reduced
and, in solution, the amount of dissociated monomer
is, therefore, strongly increased. This enhances the probability of oxidative disulde bridge formation. However,
it was important for the further course of our project
that in the new reversibly covalent “dimer” the original
contact interface surface was largely exposed. For the
rst time, we had a“pseudomonomerized” TGT structure. To increase the likelihood of the formation of this
disulde-bridged form, we later examined the enzyme
in which His 333 was also exchanged for alanine. As
aresult, the monomer fraction in solution equilibrium
increases even further.

Chapter • A Case Study: Structure-Based Inhibitor Design for tRNA-Guanine Transglycosylase
21
. Fig. 21.27 A small pocket under the loop–helix motif in the crys-
tal structure of the “pseudomonomerized” enzyme accommodates
aDMSO molecule when transferred to the cryobuffer (left). In addition, four water molecules are found in this pocket. Subsequently, ase-
What did the new form look like? In . Fig.21.26, the
dimer packing of the functional homodimer is shown
at the top, and that of the pseudomonomer spread by
the covalent disulde bridge is shown at the bottom.
The original aromatic cluster (right), in which Tyr 330
is replaced by cysteine, is rearranged to reveal the two
covalently linked cysteines in its center. The old contact
area is oriented into the crystal packing with virtually
no direct contact to any neighboring molecule. Interestingly, the loop portion of the loop–helix motif presents anew geometry. Importantly, the loop adopts an
arrangement that no longer matches the original dimer
packing of the functional homodimer for steric reasons.
Our hypothesis was that the loop also adopts this geometry in the monomeric state in solution. This idea was
supported by MD simulations. Protein crystals are briey
immersed in acryobuffer before being measured at low
temperatures. This ensures that any remaining water in
the crystals solidies into aglassy state upon freezing.
The used cryobuffer contained asmall amount of the solvent DMSO, which was not present in the crystallization
buffer. One molecule of this solvent diffused into asmall
pocket under the loop together with four water molecules
(. Fig. 21.27, top left). We took this observation as
ries of commercially available sulfoxide fragments (right) were tested
directly for their binding into the protein crystals. However, only 21.33
could be detected as ahit in the pocket
astrong indication that small molecules could be introduced as ligands into the pocket under the rearranged
loop! This was the starting point for afragment-based
lead structure search.
Awhole series of commercially available sulfoxides
was tested by directly diffusing them into the protein
crystals (. Fig. 21.27). Surprisingly, besides DMSO,
only the cyclic tetramethylene sulfoxide 21.33 was found
to be ahit. Analysis of the crystal packing of the new
hexagonal crystal form of the “pseudomonomerized”
TGT suggested that the binding pocket of the fragments
could only be reached through avery narrow solvent
channel. Therefore, only very small fragments could enter the binding pocket. However, they were not prevented
from successful binding by constrictions of the channel
in front of the binding site. Dzung Nguyen, therefore,
extended his fragment search to small analogues of 21.33
and was able to discover atotal of six additional fragments as hits. . Fig.21.28 shows the binding mode of
two of these derivatives. For 21.34, it was even possible
to accommodate two of these fragments in the pocket
at the same time. They form specic contacts with the
surrounding amino acids Gly46, Thr47, Pro56, and
Met93, which establish hydrogen bonds with the residues

. • A Fragment Opens aTransient Pocket and Suggests the Design of Bacteria-specic Inhibitors
. Fig. 21.28 Binding geometry of sulfolane 21.34, which binds anal-
ogously to sulfoxide 21.33, in the pocket below the loop–helix motif.
Asecond copy of 21.34 also ts into the pocket and displaces some of
the water molecules. The introduction of two trans-oriented hydroxy
groups allows the binding to be improved by further contacts to the
amino acids Gly6, Thr47, Pro56, and Met93. In the structure of the
of the aromatic cluster in the functional dimer. Interestingly, the sulfone group in 21.34 can be replaced by an
isosteric CF2 group to form fragment 21.35, which binds
to the protein as the R,R-stereoisomer.
The key question was whether these fragments also
bind to the enzyme in solution. To prove this, we used
NMR spectroscopy. So-called DOSY experiments were
performed, in which molecular diffusion coefcients are
compared with one-dimensional chemical shifts. For the
experiments, we used aTGT variant in which His 333
was replaced by aspartate. This variant is signicantly
weakened in its dimer stability, so that the monomer is
present in solution at appreciable concentrations. This
clearly favors the detection of the weak binding of
asmall fragment. Indeed, the binding of the racemic
21.35 was successfully detected, with abinding constant
estimated to be about 90 µM. As another fragment, for
which no crystal structure could be determined, the racemic trans-3-hydroxy-4-aminosulfolane 21.36 was detected with abinding constant in the same range.
The fragment approach described above shows that
ligands can be found that bind to the interface contact
surface of the original functional homodimer. They sta-
original functional homodimer, these residues are involved in H-bonds
to the amino acids of the aromatic cluster. The sulfone group can be
replaced by an isosteric diuoromethylene group. Fragment 21.35 and
sulfolane 21.36 have been characterized by NMR spectroscopy as micromolar binders to TGT in solution
bilize the conformationally exible loop–helix motif in
ageometry that no longer ts the original packing of the
functional homodimer. As adesign concept, it is therefore necessary to develop sufciently potent ligands for
the binding pocket under the loop–helix motif that bind
to the monomeric form of TGT. They would then have to
alter the geometry of the former interface contact surface
to such an extent that the two monomeric units would no
longer be able to recognize and interact with each other.
As aresult, the catalytically active homodimer can no
longer form and the function of the enzyme would be
blocked. With regard to the inhibition of the function of
TGT, the observed fragment structures point to afurther
strategy for the development of putative drugs.
21.16 A Fragment Opens aTransient
Pocket and Suggests the Design
of Bacteria-specific Inhibitors
Acritical aspect of any drug design project is to achieve
sufcient selectivity of the developed drug candidates.
With TGT, we have chosen an enzyme as target. Inhi-

Chapter • A Case Study: Structure-Based Inhibitor Design for tRNA-Guanine Transglycosylase
21
. Fig. 21.29 Bacterial TGT incorporates preQ1 21.2 into tRNA as
asubstrate. Instead, the human enzyme uses queuine 21.37 as the substrate at this site. On the right, complexes with three ligands are shown:
binding geometries of the parent scaffold of lin-benzoguanine 21.21
(left, light blue), of the fragment 21.38 (center, purple), and inhibitor
bition of this enzyme represents atherapeutic concept
for the treatment of Shigella dysentery. However, such
an enzyme also exists in humans. It uses queuine 21.37,
aslightly larger substrate compared to the bacterial species (. Fig.21.29). It is similar to the nal product that
bacterial enzymes produce from the precursor preQ1
21.2, which they then modify in several steps to produce
queuine (. Fig.21.1). Despite this difference in size, the
catalytic centers of the bacterial and human enzymes are
related. The human enzyme also catalyzes as adimer.
However, it is aheterodimer consisting of two different
monomeric units. This makes the contact area between
the bacterial and human enzyme signicantly different
and should simplify the selectivity problem for inhibitors
targeting the interface contact area. But what about active site inhibitors? Based on astructure determination
with afragment, Engi Hassaan observed an unexpected
result. It gave us an idea of how to selectively inhibit bac-
terial TGT.
Fragment 21.38 has amarked similarity to the amino
acid arginine. It binds to the catalytic center of TGT and
interacts with Asp 156 in amanner similar to the lin-ben-
zoguanine scaffold found in 21.21. However, the 21.38
fragment is able to open atransient pocket at this site. The
resulting pocket is lled with several water molecules. Its
opening brings the amino acids Cys 158 and Val 233,
which in the closed state are in almost van der Waals contact with each other, to agreater distance (. Fig.21.30).
The previously buried Cys 158 becomes freely accessible. With inhibitor 21.39, we have found arst candidate
that pushes into this pocket with its propargyl group and
comes close to the thiol group of Cys 158. Freely accessible thiol groups can be covalently attached to asuit-
21.39 (right, ochre, electron density outlined with amesh) extended
with apropargyl substituent are shown. Compared to 21.21, fragment
21.38 and the modied inhibitor 21.39 are able to open asmall transient pocket (marked in yellow) near Gln 203 and Gly 230
able ligand by chemical reaction. Several examples of
disulde tethering (Sect.7.10) have demonstrated this
as apromising strategy for drug design. An irreversible
covalent bond to Cys 158 with an inhibitor derived from
21.39 that binds to the active site would inhibit the function of TGT. It is interesting to note that acysteine at
this position is found only in the bacterial enzymes. The
comparable TGTs from higher developed eukaryotic organisms do not have this amino acid in that position.
This provides apromising structural site for inhibition
that is unique to bacterial enzymes: an ideal prerequisite
for the development of selective inhibitors!
21.17 Many Ways to aSmart Antibiotic
Against Shigellosis
The development of so-called “smart” anti-infectives is
particularly attractive because, when used therapeutically, they do not radically kill the entire bacterial culture
in the intestinal ora. Instead, they use specic mechanisms to prevent bacteria from becoming pathogenic.
Inhibiting TGT in Shigella could be such asuccessful
concept. The enzyme which controls the production of
the proteins that initiate the invasion process into the
epithelial cells is blocked. However, there is still along
way to go before suitable candidates are ready for clinical
testing. The goal of our academic research was to identify
potential concepts and starting points for modulating the
function of this versatile enzyme. The translation into
acommercial drug candidate for potential therapeutic
applications is still almost exclusively carried out by the
pharmaceutical industry, where the necessary infrastruc-

ab
cd
. • Many Ways to aSmart Antibiotic Against Shigellosis
. Fig. 21.30 Upper left An overlay of the binding modes of 21.21
(light blue), 21.38 (purple), and 21.39 (ocher) is shown with TGT. The
latter two ligands open atransient pocket (upper and lower right) com-
pared to 21.21 (lower left). This increases the distance between Cys 158
and Val 233. While in the complex with 21.21, the thiol group of Cys
158 is buried by the van der Waals contact with Val 233, it becomes
accessible to chemical attack by opening the transient pocket in the
complexes with fragment 21.38 and extended inhibitor 21.39. It is convenient to replace the propargyl group in 21.39 with asuitable chemi-
ture, nancial resources, and organizational structures
are available for such aproject.
The purpose of this chapter was to show, using aselected example, how the in-depth study of the properties
and functions of atarget protein can provide entry points
for manipulating its biochemical function. Several strategies can be proposed. One is the more classical block-
ing of the catalytic center. Model compounds down to
subnanomolar inhibition have been developed. Due to
their highly polar nature, they will probably have to be
used as prodrugs. To achieve specicity for bacterial enzymes only, irreversible covalent inhibition of acysteine
near the catalytic center would be conceivable. Only in
bacterial enzymes is this residue located next to atransient pocket. Ligands are needed that open the transient
pocket and bind covalently to the then accessible thiol
group of the cysteine. Since the protein binds tRNA
exclusively as a homodimer, its function can also be
disturbed by disrupting the dimer structure. On the one
hand, inhibitors of the active site can be equipped with
amolecular spike. This allows them to interfere and dis-
cally reactive group to form acovalent bond between the inhibitor and
the thiol group of Cys 158. Only bacterial TGTs contain acysteine at
this position. This is apromising starting point for the development of
selective inhibitors of bacterial TGTs. (7 https://sn.pub/AMbWJF)
rupt the contact surface between the monomer units. Ligands that bind under an exposed loop in the monomeric
state can also block dimer formation. Their binding restructures the original monomer units in away so that
they no longer t together for dimer formation. Since
many enzymes have feedback regulatory mechanisms, the
ligand-induced transformation of TGT into an inactive
“resting state” may provide another strategy. By forming
an alternatively packed dimer, the enzyme is unable to
bind its substrate tRNA. Substances that stabilize this
“dormant” state may, therefore, also inhibit the function
of the target enzyme.
There seem to be many approaches to the desired goal.
So, does TGT represent aspecial case that has led to so
many different approaches to inhibiting it? Probably not,
because proteins are “social,” involved in many networks
and, thus, participate in many mechanisms. It is more
aquestion of time and thoroughness of characterization,
regarding how many possible inhibitory mechanisms can
be traced. In terms of drug design, TGT is aprime example. You can see how many different concepts are open to

Chapter • A Case Study: Structure-Based Inhibitor Design for tRNA-Guanine Transglycosylase
21
drug designers if they only “get to the heart” of the target
structure for along enough time! In the end, it has to be
shown pharmacologically which of these concepts is the
most suitable for therapeutic success.
21.18 Synopsis
Shigella dysentery is asevere bacterial diarrheal ill-
-
ness. Shigella bacteria that are ingested with contam-
inated water or food adhere to epithelial cells in the
intestinal mucosa. To gain entrance to these cells,
the bacteria produce their own virulence factors, socalled invasins.
The genes encoding the invasins will only be tran-
-
scribed in sufcient amounts if the transcription
factor VirF is present in sufcient quantities. Aprerequisite for its efcient protein biosynthesis is the
tRNA-modifying enzyme tRNA-guanine transglycosylase, which catalyzes the incorporation of the modied preQ1 base into the wobble position of certain
tRNAs.
Afunctional assay recording the exchange of gua-
-
nine by radioactively labeled guanine can determine
the potency of ligands inhibiting the function of the
target enzyme.
The rst hits were detected by using the de novo design
-
program LUDI, and the predicted binding mode of
amicromolar hit was conrmed by crystallography.
The active site shows adaptations by ipping apep-
-
tide bond and mediating important interactions to
the substrates through awater molecule.
Virtual screening suggests abroad variety of basic
-
scaffolds for inhibitor design. Alin-benzoguanine
scaffold served as the most promising lead structure.
Substitutions at the 2- and 4-positions of the lin-ben-
-
zoguanine scaffold lead to very different increases in
afnity. The addition of a2-amino group leads to the
presence of two basic centers in the molecule. Interestingly, the less basic center is protonated and, thus,
charged by the local polarity in the binding pocket.
This leads to salt bridge-like interactions with two
aspartate residues. Nevertheless, the addition of the
2-amino group also results in an enormous increase in
afnity. Here, the surprising increase in afnity can be
attributed to the formation of two parallel H-bonds,
which do not experience secondary repulsive effects
and, thus, turn out to be very strong.
Substitutions at the 4-position have to interfere with
-
and partially replace acontiguous water network between two facing aspartic acids. They can potentially
link the parent scaffold with substituents which ll
asmall hydrophobic pocket. Asignicant potency enhancement can be achieved only if the spacer linking
the two portions contains polar atoms to cross the
water network. These atoms can actively participate
in the network.
Several iterative cycles of design, crystal structure
-
analyses, and inhibitor syntheses were required to
develop the initial double-digit micromolar hits
into subnanomolar inhibitors. The best potency is
achieved by lin-benzoguanines carrying both a2- and
4-substituent.
Surprisingly, the enzyme is only functional as aho-
-
modimer. Thus, disruption of the dimer geometry
may provide afurther principle for inhibitor design.
Computational analysis and subsequent targeted mu-
tagenesis of individual residues in the dimer interface
identied residues that contribute primarily to the sta-
bility of the contact area expanding over more than
1600 Å2. H-bonds and acluster of aromatic amino
acids were found to be critical. The cluster is embed-
ded in an environment of hydrophobic amino acids
and prevents the entry of water molecules.
Since the interfacial contact region is located near the
-
catalytic center, active site inhibitors with very long,
spike-like substituents were able to disrupt the dimer
contact and convert the dimer, at least partially, to
amonomer.
With ligands carrying bulkier, partially sugar-con-
-
taining substituents in the 4-position, anew crystal
form was found by chance under the same crystalli-
zation conditions. The enzyme appears with atwisted
arrangement of the structurally unchanged mono-
mer units. Anew interface contact area of almost
the same size is formed. In this ligand-induced form,
the monomer units pack together in such away that
the tRNA substrate can no longer be bound. Thus,
ligands that induce and stabilize this twisted, catalyt-
ically inactive form are another concept for the devel-
opment of inhibitors for the enzyme.
Since crystallography only characterizes the end-
-
points of apossible dynamic rearrangement, it was
possible to infer adynamic transformation in solution
by spin labeling using ESR spectroscopy. By introduc-
ing 19F-labeled tryptophan, NMR spectroscopy was
also able to detect the ligand-induced dynamic rear-
rangement process between the catalytically compe-
tent and the twisted inactive dimer forms in solution.
One ligand proves to be an effective stabilizer of the
twisted inactive form and can, therefore, be consid-
ered as astarting point for an alternative inhibitor
design.
Surprisingly, acysteine residue introduced into the
-
interface contact surface by mutagenesis led to the
formation of a new crystal form under oxidative
conditions, in which the former contact surface is
structurally exposed by the altered dimer packing.
The new packing was created by adisulde bridge
between the cysteine residues introduced into the di-
mer. The enzyme, thus, assumes a“pseudomonom-
Соседние файлы в папке Библиотека им академика М.И. Перельмана
