Proteinprotein and proteinligand interactions

1
Protein-protein and protein-ligand
interactions (NMR in drug discovery)
2
Protein-protein interactions
General background on protein-protein
interactions Russell RB, Alber F, Aloy P, Davis
FP, Korkin D, Pichaud M, Topf M, Sali A. A
structural perspective on protein-protein
interactions. Curr Opin Struct Biol. 2004
14313-324. Review Zuiderweg ER. Mapping
protein-protein interactions in solution by NMR
spectroscopy. Biochemistry. 2002
411-7. Techniques Otting G, Wüthrich K.
Heteronuclear filters in two-dimensional
1H,1H-NMR spectroscopy combined use with
isotope labelling for studies of macromolecular
conformation and intermolecular interactions. Q
Rev Biophys. 1990 2339-96. Takahashi H,
Nakanishi T, Kami K, Arata Y, Shimada I. A novel
NMR method for determining the interfaces of
large protein-protein complexes. Nat Struct Biol.
2000 7220-223. Gaponenko V, Altieri AS, Li J,
Byrd RA. Breaking symmetry in the structure
determination of (large) symmetric protein
dimers. J Biomol NMR. 2002 24143-8. Leonov A,
Voigt B, Rodriguez-Castaneda F, Sakhaii P,
Griesinger C. Convenient Synthesis of
Multifunctional EDTA-Based Chiral Metal Chelates
Substituted with an S-Mesylcysteine. Chemistry.
2005 Mar 30 Epub ahead of print PMID
15798974 Dosset P, Hus JC, Marion D, Blackledge
M. A novel interactive tool for rigid-body
modeling of multi-domain macromolecules using
residual dipolar couplings. J Biomol NMR. 2001
20223-231. Scherf T, Hiller R, Anglister J.
NMR observation of interactions in the combining
site region of an antibody using a spin-labeled
peptide antigen and NOESY difference
spectroscopy. FASEB J. 1995 9120-126. Riek R,
Fiaux J, Bertelsen EB, Horwich AL, Wüthrich K.
Solution NMR techniques for large molecular and
supramolecular structures. J Am Chem Soc. 2002
12412144-12153. Fiaux J, Bertelsen EB, Horwich
AL, Wuthrich K. NMR analysis of a 900K GroEL
GroES complex. Nature. 2002 418207-211. Sakakura
M, Noba S, Luchette PA, Shimada I, Prosser RS. An
NMR Method for the Determination of
Protein-Binding Interfaces Using Dioxygen-Induced
Spin-Lattice Relaxation Enhancement. J Am Chem
Soc. 2005 1275826-5832.
3
Ways of identifying proteinprotein interactions
EM image reconstruction
Electron tomography
Y2H screen
X-ray
NMR
This review reports that there are 12,000 known
structures of assemblies with proteinprotein
interactions
Russell et al. Curr Opin Struct Biol. 2004
14313-24.
4
Pros and cons of using NMR to investigate
complexes

• Cons
• Cant use NMR to solve structures of large
  complexes such as virus particles or ribosomes
• Approach becomes tedious for larger complexes
• Pros
• Can investigate weak complexes that do not
  crystallize well
• Can determine equilibrium constants and on and
  off rates
• Can map interaction faces
• Can detect conformational changes that accompany
  binding

5
Proteinprotein interactions present a major
bottleneck for structural genomics (SG)

• Oligomers of identical subunits are frequently
  found
• If they contain a large number of subunits, they
  usually are not suitable for high-throughput NMR
• Proteins consisting of different subunits were
  largely ignored in the first rounds of SG
• Subunits that fail to fold on their own (those
  requiring a second chain) have been ignored
• These challenges likely will be addressed in PSI-3

Example 104 aa hexamer, HSQC- Structure solved
by X-ray 1VK0
6
800 kD 12-mer protein GroEL
Heteronuclear 2D 15N,1H-correlation spectra of
the uniformly 15N,2H-labeled 800-kDa tetradecamer
protein GroEL from E. coli. (a) 2D 15N,1H-TROSY
spectrum, measuring time 20 h, acquired data
size 256 1024 complex points, t1,max 32 ms,
t2,max 100 ms. (b) 2D 15N,1H-CRIPT-TROSY
spectrum, measuring time 20 h, acquired data
size 100 1024 complex points, t1,max 10 ms,
t2,max 100 ms.
Riek et al. J. Am. Chem. Soc. 12412144
7
Ways of identifying proteinprotein interactions
by NMR

• Nuclear Overhauser effect
• RDCs
• Chemical shift perturbation mapping
• Cross-saturation
• Titration studies
• Mapping with dynamics
• Mapping with amide-proton exchange
• Mapping with paramagnetics
• Use of dipolar couplings to determine relative
  orientations of subunits
• Orientational constraints from differential
  anisotropic 15N relaxation in protein subunits

Zuiderweg ER. Biochemistry 2002 411-7
8
Nuclear Overhauser effect
Full three-dimensional structure of the complex
is determined Method is only applicable to tight
complexes (Kd lt 10 ?M) Can be used with
isotope-editing/filtering (differentially labeled
subunits)
9
Intrasubunit vs. intersubunit NOEs and RDCs
confirm dimeric structure

• NMR results for the At5g22580 homodimer.
• Strips from a 3D 13C NOESY spectrum recorded at
  750 MHz illustrating intersubunit (red) and
  intrasubunit (black) NOEs.
• Correlation between 105 measured 1DNH residual
  dipolar couplings and those calculated from the
  lowest energy monomeric structure.
• Similar correlation between the measured residual
  dipolar couplings and those calculated from the
  lowest energy dimeric structure.

Cornilescu et al. (2004) J. Biomol. NMR 29387
10
Structure of the homodimer
Cornilescu et al. (2004) J. Biomol. NMR 29387
11
Chemical shift perturbation mapping
Most widely used NMR method to map protein
interfaces Usually monitored by 1H-15N
HSQC Interfaces usually map to residues showing
the largest changes Beware of conformational
effects If more resonances change than those in
the expected contiguous interaction surface,
suspect an allosteric process Can be used with
large systems eg 51 kDa complex of FimC-adhesin
and FimH
12
Cross-saturation
Low-resolution identification of the
interface Uses the same physical processes as the
NOE experiment Donor protein is unlabeled
observed protein is 2H/15N-labeled with amide
deuterons are exchanged back to protons NMR
experiment starts with steady-state saturation of
aliphatic proton resonances (all on donating
protein) Saturation is carried by
cross-relaxation from the donor to the amide
protons of the acceptor. Detection is by 1H-15N
HSQC or, for larger proteins, 1H-15N TROSY
(monitor change in intensity without and with
saturation) Labeling can be reversed to obtain
the other interface Can be used with very large
donor proteins and to study weaker complexes
13
Mapping of protein-protein interface by
transferred cross-saturation experiment

• Sample requirement
• Perdeuteration
• Solvent (90D2O,10H2O)

Takahashi et al., Nature Struct. Biol. (2000), 7,
220
14
OMTKY3-Chymotrypsin, a model system for
protein-protein interactions
15
1H,15N-HSQC of (U-2H, U-15N) OMTKY3 in complex
with chymotrypsin in 90D2O / 10H2O
Jikui Song PhD Thesis
16
Effect of the saturation time on the intensity
ratios of the NH crosspeaks with irradiation to
those without irradiation
Jikui Song PhD thesis
17
Chemical shift perturbation vs. transferred
cross-saturation
Saturation Time 1.2s
Jikui Song unpublished results
18
Chemical shift perturbation vs. transferred
cross-saturation
Chemical shift perturbation Red, ??(HN)gt0.2 ppm
Orange, ??(HN)gt0.1 ppm
Transferred cross-saturation Red, Intensity
ratiolt0.6 Orange, Intensitylt0.75
X-ray (lt4Å contact) Red, 7-30 contacts Orange,
1-3 contacts
Jikui Song unpublished results
19
Titration studies
Provide estimates of the affinity, stoichiometry,
and specificity of binding as well as the
kinetics of binding Single set of peaks that
move weaker interaction (Kd lt 10 ?M) can extract
binding constant from fitting the titration
curve Separate signals for free and bound forms
tighter interaction need to cross assign signals
from free and bound binding constant can be
extracted from analysis of peak heights In
general Kd values in the range of 10-6 to 10-2 M
can be measured Separate binding steps can be
differentiated if their Kd values differ by 10 or
more
20
Protein-protein binding site mapped by chemical
shift change on titrating in a peptide fragment
Jikui Song et al. (2005) J. Mol. Biol.
3541043-1051
21
Model for the localization of the dimeric
Roadblock protein and interaction with IC74
22
Mapping with dynamics
In some cases, dynamics becomes quenched at a
proteinprotein interface in others,
not Interesting results but not diagnostic for
identifying interfaces
23
Mapping with amide-proton exchange
The exchange rate at proteinprotein interface
decreases Often the interaction decreases
exchange throughout the protein(s) It can be
dangerous to use this approach to identify the
interface
24
Mapping with paramagnetics
Water soluble paramagnetic species Paramagnetic
gadolinium-EDTA relaxes surface residues Residues
in proteinprotein contact areas should be
shielded from this effect Covalently attached
paramagnetic species (chelated metal or spin
label) used to map intermolecular contacts or
break magnetic symmetry with homodimer Griesinger
has designed a good chelating reagent Pseudocontac
t shifts can be used
25
Use of paramagnetic probe to break symmetry in a
homodimer
Site-directed paramagnetic labeling was
achieved by modification of the cysteine residue
107 with S-(2-pyridylthio)-cysteaminyl-EDTA
(Toronto Research Chemicals).
15N TROSY spectrum collected at 900 MHz on 2H,
15N, 13C enriched ILV labeled sample of
STAT4NT-EDTA-Co2. Spectral widths were 16 kHz
and 3.2 kHz in the direct and indirect
dimensions, respectively, and the number of t1
increments was 128. The insert in the upper left
corner shows an expanded region corresponding to
the boxed area in the middle of the spectrum. The
peaks belonging to the paramagnetic species are
colored red and green to represent the intra- and
intermonomer pseudocontact shifts, respectively.
Gaponenko V et al. (2002) J Biomol NMR 24143
26
Use of paramagnetic probe to break symmetry in a
homodimer
Correlation between observed and calculated PCSs
for HN protons in the STAT4NT-EDTA-Co2 dimer.
The calculated PCS values were obtained from the
dimer structure using XPLOR-NIH. Intramonomer
PCSs are shown in red and intermonomer PCSs are
shown in green. The linear correlation
coefficient is 0.99.
Gaponenko V et al. (2002) J Biomol NMR 24143
27
Use of oxygen as a surface probe dioxygen-induced
spin-lattice relaxation enhancement
The method of comparing oxygen-induced
spin-lattice relaxation rates of free protein and
protein-protein complexes, to detect binding
interfaces, offers greater sensitivity than
chemical shift perturbation. Not necessary to
heavily deuterate the labeled protein, as is the
case of cross saturation experiments.
M. Sakakura M et al. (2005) J. Am. Chem. Soc.
1278526
28
Dioxygen-induced spin-lattice relaxation
enhancement FBFc complex vs. FB protein alone
Depiction of R1P(complex)/R1P(free) mapped onto
FB. A gradation of red (strongest binding or
greatest change of O2 accessibility upon binding)
to white (no binding) is assigned to residues
depending on the magnitude of R1P(complex)/R1P(fre
e).
M. Sakakura M et al. (2005) J. Am. Chem. Soc.
1278526
29
Use of dipolar couplings to determine relative
orientations of subunits
Can use differential labeling Can be used to dock
subunits with sparse number of NOE or other
constraints
Orientational constraints from differential
anisotropic 15N relaxation in protein subunits
30
Prediction of proteinprotein interactions and
binding sites

• Most promising approach to the problem will be to
  use bioinformatics followed up by high-throughput
  screening
• Different approaches are being used
• Subunits with sequence similarity to single
  domains in multi-domain proteins search for
  sequence representing other isolated domains
• Look for genes that tend to be located next to
  one another in genomes
• Search the literature for evidence for oligomers
  that can be linked to ORFs
• Sometimes the structure of one subunit can be
  analyzed by structural similarity searches to
  reveal a potential protein binding site

31
Protein-ligand interactions Reviews Takeuchi K,
Wagner G. NMR studies of protein interactions.
Curr Opin Struct Biol. 2006 16(1)109-17. Carlomag
no T. Ligand-target interactions what can we
learn from NMR? Annu Rev Biophys Biomol Struct.
2005 34245-66. Lymphotactin-ligand
interactions Peterson FC, Elgin ES, Nelson TJ,
Zhang F, Hoeger TJ, Linhardt RJ, Volkman BF.
Identification and characterization of a
glycosaminoglycan recognition element of the C
chemokine lymphotactin. J Biol Chem. 2004 Mar
26279(13)12598-604. NMR in drug discovery and
drug design Hajduk PJ, Greer J. A decade of
fragment-based drug design strategic advances
and lessons learned. Nat Rev Drug Discov. 2007
6(3)211-9. Hajduk PJ, Meadows RP, Fesik SW.
Discovering high-affinity ligands for proteins.
Science. 1997 Oct 17278(5337)497,499. Shuker
SB, Hajduk PJ, Meadows RP, Fesik SW. Discovering
high-affinity ligands for proteins SAR by NMR.
Science. 1996 Nov 29274(5292)1531-4. Moore JM.
NMR screening in drug discovery. Curr Opin
Biotechnol. 1999 Feb10(1)54-8. Review. Fejzo J,
Lepre CA, Peng JW, Bemis GW, Ajay, Murcko MA,
Moore JM. The SHAPES strategy an NMR-based
approach for lead generation in drug discovery.
Chem Biol. 1999 Oct6(10)755-69. PMID 10508679
UI 99439951 Reibarkh M, Malia TJ, Hopkins BT,
Wagner G. Identification of individual
protein-ligand NOEs in the limit of intermediate
exchange. J Biomol NMR. 2006 Sep36(1)1-11.
32
NMR methods for studying protein-ligand
interactions

• Sample properties that need to be considered
• Binding affinity
• tight binding (koff lt 105 s-1)
• weak binding (koff gt 104 s-1)
• Sizes of the protein and ligand
• Small enough to be observable directly by NMR
• Too big to be observed directly by NMR

33
Tight binding (koff lt 102 s-1)

• At partial saturation observe separate signals
  from free and bound
• Monitor signals from the complex itself
• Differentially label the protein and / or the
  ligand
• Can detect
• Chemical shifts
• NOEs (intra- and inter-molecular)
• RDCs
• Approach is limited to complexes that are small
  enough to yield resolved NMR signals
• Can decrease linewidths by deuterium labeling

34
Weak binding (koff gt 104 s-1)

• Large protein (gt50 kDa) or small protein (lt50
  kDa)
• Can monitor ligand(s)
• Saturation transfer difference (STD)
• Diffusion affected by binding
• Relaxation affected by binding
• Transferred NOEs
• CCR rate (cross correlated relaxation)
• Solvent saturation transfer (WaterLOGSY)
• Relaxation transferred from spin-labeled protein
• Small protein (lt50 kDa)
• Can monitor the protein
• Chemical shifts of a labeled protein
• Smaller protein U-15N, U-13C
• Larger protein 13C methyl labeling

35
Example human lymphotactin (hLtn)
Solution structure
Kuloglu et al. (2001) Biochemistry, 40,
12486-12496 (2001)
36
Interaction of hLtn with heparin

• hLtn interacts specifically with cell surface
  glycosaminoglycans
• 15N-labeled hLtn was titrated with
  heparin-derived oligosaccharides
• Mutants were designed to test the heparin binding
  mechanism deduced from NMR (affected residues
  conserved Lys and Arg)
• 18 43
• hLtn ...CVSLTTQRLPVSR...VIFITKRGLKVCAD...
• hMIP-1a ..CCFSYTARKLPRNF...VVFQTKRSKQVCAD...

37
Titration of 15N-hLtn with heparin-
pentasaccharide

• Residues perturbed
• R23, R43, R70
• K42, K46
• V21, I24, T41,
• L45, S62

38
Mapping the heparin-pentasaccharide binding
surface of hLtn
39
Heparin-pentasaccharide binding site of hLtn
Mutations R23A or R43A abolished binding
40
Summary heparin-pentasaccharide binding to hLtn
a. 2D HSQC spectra of hLtn (125 µM) in the
presence (green contours) and absence (orange
contours) of a 3-fold molar excess of synthetic
heparin pentasaccharide. b. Arg side chain NH
signals from HSQC spectra of hLtn with arrows
highlighting the selective perturbation of Arg-23
and Arg-43 in response to pentasaccharide
binding. c. Titration of hLtn with
pentasaccharide shifts a unique set of resonances
in a concentration-dependent manner. Orange,
magenta, blue, and green contours correspond to
0, 125, 250, and 375 µM heparin. d. Combined
backbone 1H and 15N chemical shift perturbations.
Lysine and arginine residues analyzed by
mutagenesis are highlighted (blue bars). Other
basic residues are indicated with dashed bars. e.
Ribbon diagram of hLtn showing the location of a
heparin binding surface.
Peterson, F. C. et al. J. Biol. Chem.
200427912598-12604
41
Drug discovery and development
Problems with conventional methods for
high-throughput screening of large libraries of
compounds Extensive development required False
positives from artifacts of the detection
method Applicable to limited classes of
molecules Ideal assay system Simple to
implement Composed of only a few
components Amenable to all classes of
molecules Adaptable to large libraries
42
Fragment based screening
Nat Rev Drug Discov. 2007 6(3)211
43
Potent inhibitors (IC50 lt 100 nM) derived from
experimentally driven fragment-based screening
and design
Nat Rev Drug Discov. 2007 6(3)211
44
NMR-based screens for drug discovery and
development

• Method based on observation of the protein
• SAR by NMR (Abbott Labs)
• Screen 15N labeled target protein against small
  molecules detect altered peak positions
• Methods based on observation of small molecules
• SHAPES (Vertex)
• Monitor signals from small molecules infer
  binding from line broadening or transferred NOE
• Affinity NMR, DECODES (Novartis)
• Infer binding from change in diffusion in
  presence of macromolecule to which it binds

45
SAR by NMR
Science 2741531 (1996)
46
SAR by NMR an example
Superposition of 15N-HSQC spectra for FKBP in the
absence (magenta contours) and presence (black
contours) of compound 3. Both spectra were
acquired in the presence of saturating amounts of
2 (2.0 mM). Significant chemical shifts changes
are observed for labeled residues.
Science 2741531 (1996)
47
SAR by NMR another example
Shown are matrix metalloproteinase 3 (MMP3) (a)
and BCL-XL proteins (b). In each case at the top,
the identified fragment leads are shown with cyan
carbons, whereas the linked compounds are denoted
with green carbon atoms. All structures were
experimentally determined by NMR. The chemical
structures (and in vitro potencies) of the
fragment leads and subsequent high-affinity
linked compounds are shown in the lower part of
the figure. NMR, nuclear magnetic resonance SAR,
structureactivity relationships.
48
Screening methods based on protein chemical shift
changes (SAR by NMR)
Advantages Simple assay 15N labeling guarantees
no background signals Can differentiate different
binding sites on the basis of peak
patterns Disadvantages Number of compounds
screened at a time is limited by the sensitivity
(can overcome to some extent by using
cryoprobe) Method is not amenable to very large
proteins (can offset by using selective labeling)
49
SAR by NMR
In the first step, a library of small molecules
is screened for binding to a protein. Binding is
detected from the amide chemical shift changes
observed in 2D HSQC spectra. Once two ligands are
identified that bind to the protein, the
structure of the ternary complex is determined
(middle left).
Science (1997) 278 497-499
50
SAR by NMR
Comparison of different building-block
approaches. With combinatorial chemistry, many
linked compounds (fragment 1) (fragment 2)
(different linkers) are synthesized that contain
all combinations of fragments and linkers. In
contrast, with SAR by NMR, only a few compounds
need to be synthesized (yellow highlighted boxes)
because the fragments that bind to the protein
are identified before linking (arrows), and the
linkers are selected on the basis of structural
information.
Science (1997) 278 497-499
51
Cryoprobe technology for SAR by NMR
Sensitivity-enhanced 1H/15N HSQC spectra acquired
in 10 min on a 50 ?M sample of the catalytic
domain of stromelysin using (A) a dual (15N/1H)
inverse CryoProbe (Bruker) equipped with a lock
and a z-gradient and (B) a conventional
triple-resonance (13C/15N/1H) inverse probe
(Nalorac). Based on peak volumes, an average gain
in signal-to-noise of 2.4 was realized with the
CryoProbe as compared to the conventional TXI
probe.
Hajduk et al. (1999) J Med Chem. 422315-2317
52
Expansions of 1H/15N HSQC spectra obtained with a
CryoProbe on 50 ?M samples of stromelysin (A) in
the absence (black contours) and presence (red
contours) of a mixture of 100 known nonbinders
and (B) in the absence (black contours) and
presence (red contours) of a mixture of 100 known
nonbinders plus 3-4-(4-cyanophenyl)phenoxypropan
ohydroxamic acid. Compounds were tested at a
concentration of 50 ?M each.
Hajduk et al. (1999) J Med Chem. 422315-2317
53
Theoretical occupancy levels of a target protein
as a function of ligand KD under the conditions
of 0.5 (thick dotted line) and 0.05 (thick solid
line) mM ligand and protein. The detection limit
for ligands in fast exchange (thin solid line) is
set at 20 of full occupancy.
Hajduk et al. (1999) J Med Chem. 422315-2317
54
Make 100 - 200 mg batches of 15N proteins
Screen 100 compounds / protein Can screen
100,000 compounds / week Cost effective vs other
high-throughput screening approaches e.g.
fluorescence or radiolabeling For larger
proteins, replace uniform 15N labeling with I,L,V
13C labeling (Gardener et al., JACS 120, 1173
(1998)
Steve Fesik ENC 4/14/00 19 of leads at Abbott
Labs. are coming from NMR screening
/ gram cell paste 15N 4.50 13C 57.00
55
SHAPES approach (Vertex)
SHAPES strategy uses standard one-dimensional
(1D) line broadening and 2D transferred nuclear
Overhauser effect (tNOE) measurements to detect
binding of a limited (200) but diverse library
of low molecular weight, soluble scaffolds to a
potential drug target.
J. Moore (1999) Biopolymers 51221-43
56
SHAPES approach
1D 1H spectra of (bottom) free ligand 1 and (top)
the ligand 1 in the presence of IMP
dehydrogenase, a 224 kDa protein. The significant
line broadening observed in the presence of the
enzyme indicates binding of the small molecule.
Line widths at half height for the
furthest downfield component of 1 (leftmost peak)
are 3 Hz (bottom) and 30 Hz (top).
J. Moore (1999) Biopolymers 51221-43
57
SHAPES approach
1H spectra of (top) a mixture of two ligands
compared to (bottom) the mixture of the ligands
in the presence of p38 MAP kinase, 40 kDa
protein. Resonances from nicotinic acid (top left
structure) and 2-phenoxy benzoic acid (top right
structure) are marked with solid and dashed
arrows, respectively. The peak at 7.2 ppm
consists of overlapping resonances from both
compounds.
J. Moore (1999) Biopolymers 51221-43
58
SHAPES approach
Example showing how fusion of fragments with a
common scaffold can lead to a potent inhibitor.
Dissociation constants of the compounds for p38
as determined by NMR diffusion measurements are
given below each compound. Dissociation constants
for compounds marked with asterisks could not be
determined using nmr diffusion methods. The value
given for the compound shown in green at the
lower left was estimated from line-broadening and
transferred NOE data, and the value shown for the
compound in magenta at the lower right was
determined enzymatically.
J. Moore (1999) Biopolymers 51221-43
59
Labeling strategies for determining NMR
structures of complexes once a lead has been found
J. Moore (1999) Biopolymers 51221-43
60
Transferred NOE approach to investigating
molecular interactions
Use opposite sign of the NOE to provide
selectivity Since the ligand is observed, the
method is independent of the molecular weight of
the receptor (can be used for very large
proteins) Can combine with other approaches
TOCSY, COSY, HMQC (1D variants of 2D and 3D
experiments) No need to separate the mixture
binding compound can be determined from its
spectral properties
61
(No Transcript)
62
Diffusion methods for screening
DECODES (Diffusion EnCODEd Spectroscopy) TOCSY /
DOSY Use bipolar pulses to avoid exchange
artifacts Results give binding affinity in rank
order more tightly binding molecules diffuse
more like the macromolecule Uses very little
protein 2 ?M concentration Can screen 10
compounds at a time in 15 min (deconvolute
spectra) Doesnt require labeling of the
macromolecule (however, doesnt identify where
the ligand binds)