Interactions understood. Leads improved.
From fragments to transition state analogues –
drug discovery using SPR biosensor technology
Helena Danielson
Uppsala, Sweden
Proteinase 2013
8th RSC / SCI symposium on Proteinase Inhibitor Design
15th to 16th April 2013 at Novartis, Basel, Switzerland
Protease case studies
• Fragment screening – MMP-12 and HIV-1 protease
• Chemodynamics – BACE-1
• Kinetics and thermodynamics – thrombin
• Slow tight-binding inhibitors – renin & MMP-12
Screening fragment libraries against proteases
MMP-12
and
HIV-1 protease
1st SPR-based fragment screen – MMP-12
Zinc metallo proteinase
Challenging target for SPR due to:
• Autoproteolysis
• Dependence on high [Ca2+]
Screening design, reference compounds and reference surfaces
• The tools used
Fast dissociating
reference compound (RC)
Ki= 50 nM
Slow dissociating
reference compound
(ilomastat) Ki= 4 nM
Sensor surface validation and quality
wt MMP-12
+20 μM RC
Inactive MMP-12 mutant
+20 μM RC
Ilomastat stabilized MMP-12
+20 μM RC
Carbonic anhydrase
+20 μM furosemide
2-step screening process
Screening
Characterization
Kinetic analysis
wt
MMP-12
stabilized
MMP-12
mutant
MMP-12
Carbonic
anhydrase
Verifying the hits – inhibition analysis
B7
B7
IC50=0.29 µM
B7 is a good competitive inhibitor!
B6 less efficient
Crystallography confirms B7 as an active site
ligand
IC50=355 µM
Conclusions from MMP-12 fragment screen
Proof-of-principle!
• SPR can be used for fragment screening
• … even for challenging protease targets
Let’s try another challenging protease….
Screening for fragments binding wt & resistant HIV-1 protease
• Wild type
• G48V
• V82A
• I84V
Typical fragment sensorgrams
No saturation for
concentration series
No kinetic information
BEA00564
RU
10
81.8
8
Response
6
4
2
0
-2
-4
-6
-10
0
0
-5
0
5
10
15
Tim e
20
25
30
35
s
300
Detection of weakly interacting specific ligands
Normalized response (RU)
Steady-state analysis of sensorgrams
KD = 50 µM
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
0
0
1
0 7 14 21 28 35 42 49 56 63 70
Time
(s)
0
200
400
Concentration
(µM)
KD = 500 µM
0 7 14 21 28 35 42 49 56 63 70
Time
(s)
Separation of “specific” from “unspecific” interaction
Specific = A single “high affinity” interaction
Unspecific = one or several lower affinity interactions
A
2
Normalized response (RU)
1.6
ed
erv
s
Ob
i ng
d
n
bi
1.2
ding
Specific bin
0.8
in
cb
i
f
i
ec
din
g
p
Uns
0.4
0
0
10
20
Concentration (µM)
B
2
)
0 mM NaCl
300 mM NaCl
30
40
Overcoming challenges by extended data analysis
BEA00564
RU
30
8
25
6
14
12
10
2
0
8
15
Res ponse
4
10
5
-2
-6
-10
-5
0
5
10
15
20
25
30
35
81.8
-5
-10
-5
0
5
10
15
20
25
30
s
Tim e
300
-5
0
5
10
15
20
25
30
0
0
300
Hit selection criteria
• Low degree of un-specific interaction?
• High signal at saturation (or apparent saturation)?
• High signal at highest concentration?
0
35
s
Tim e
102
0
0
2
-4
-10
35
197
0
4
-2
s
Tim e
6
0
0
-4
BEA01625
RU
20
Res ponse
Res ponse
BEA01381
RU
10
300
Succeeding with proteases in SPR-based fragment screening
• Essential tools
– Reference surfaces
– Reference compounds
– Orthogonal assay
• Minimize autoproteolysis by
– Optimization of conditions
• Suboptimal pH for catalysis
• High Ca2+-concentration (for MMP-12)
– Optimization of sensor surfaces
• Low immobilization levels
• Crosslinking
• Engineer protein for increased stability
– Experimental design
• Competition mode (fast dissociating reference compound)
• Block active site (slow dissociating reference compound)
Chemodynamics – Identifying dominating interactions
and defining relevant experimental conditions
BACE-1
BACE1: β-secretase, β-site amyloid precursor protein cleaving
enzyme
• BACE1 (alias Asp2 or memapsin 2)
Membrane
spanning region
BACE-1 function and localization
Amyloidogenic APP processing
Siegenthaler BM, Rajendran L.
Neurodegener Dis. 2012;10(1-4):116-21.
Characterization of BACE inhibitors – standard procedures
No discernible correlation
between BACE inhibition and APP
cleavage in cells
Why different?
• Ectodomain vs. full length
enzyme?
• Conditions (pH 4.5 vs. 7.4)
IC50 (nM)
Compound
Enzyme
assaya
Cell
assayb
1
8
0.1
2
29
18
3
9
86
4
13
3
5
5
900
6
30
?
• Ca2+?
Reference: Domínguez, et al Effect of protonation state of the titrable residues on the
inhibitor affinity to BACE1. Biochemistry, 2010; 49, 7255–7263.
Design of sensor surfaces for BACE1
• Full length BACE1
immobilized in a lipid
membrane via antibody
capture
• Ectodomain BACE1
immobilized directly (no
transmembrane region)
A surface plasmon resonance based biosensor with full-length BACE1 in a reconstituted
membrane. Christopeit, T., et al. Analytical Biochemistry 2011, 414 pp. 14-22.
Inhibitor interactions with BACE1
Same KD for truncated and
full length BACE at pH 4.5 (for
all compounds)
Different KD for truncated
and full length BACE at pH
7.4 (for some compounds)
Different KD at pH 4.5 and 7.4
for both truncated and full
length BACE (for some
compounds)
What is relevant?
Inhibitor interactions with BACE1
Cell assay
IC50 (nM)
0.1
18
86
3
900
Correlation between IC50 and KD at 7.4!
pH effect understood via modelling of protonation of inhibitors and Asp dyad
Interpreting the data
Hypothesis
Inhibitors bind BACE1 at the cell
surface (neutral pH)
BACE1 is internalized into
endosomes for cleavage (acidic pH)
Inhibitors need to bind BACE1 at
neutral and acidic pH!
Siegenthaler BM, Rajendran L.
Neurodegener Dis. 2012;10(1-4):116-21.
Effect of protonation state of the titrable residues on the inhibitor affinity to BACE1.
Domínguez, et al Biochemistry, 2010; 49, 7255–7263.
Conclusions – the use of chemodynamic analysis
• SPR analysis of interactions is a good complement to
inhibition analysis, conditions do not have to be optimized for
catalytic efficiency
• Varying the experimental conditions is favourable for
– improved understanding of the interaction
– designing a relevant assay for selection of hits/leads
Correlating kinetics and thermodynamics with
structure
Thrombin
Melagatran and P3 analogues
Temperature dependence of kinetics
5
Melagatran
1
2
3
4
5
15
25
35
45 °C
Temperature dependence of kinetics
45 °C
kon [M-1s-1]
5 °C
koff [s-1]
Kinetic profiling at different temperatures by SPR
45 °C
5 °C
Temperature dependence of kon, koff, and KD
Thermodynamic profiles
Association
30
kJ/mol
25
15
10
5
0
Mel
1
2
3
4
80
70
60
50
40
30
20
10
0
5
30
kJ/mol
kJ/mol
25
20
15
10
5
0
Mel
1
2
3
4
-15
-35
-55
1
2
3
4
5
80
70
60
50
40
30
20
10
0
5
-25
-45
Mel
35
SPR
-5
Mel
1
2
3
4
5
Mel
1
2
3
4
5
Mel
1
2
3
4
5
-5
-15
kJ/mol
20
Equilibrium
kJ/mol
35
kJ/mol
Stopped
flow
Dissociation
-25
-35
-45
-55
Mel
1
2
3
4
5
0
• Melagatran is the most enthalpy-driven interaction.
• Largest thermodynamic differences in association
phase
• Smallest effects at equilibrium
ΔH
TΔS
ΔG
-10
kJ/mol
ITC
-20
-30
-40
-50
Correlation between structure, kinetics and thermodynamics
• kon increased with lipophilicity of the inhibitors
– Increased enthalpic component
– Corresponding decreased entropic component
• koff decreased with increased chain length
– Smaller increases and decreases of enthalpic and entropic
components
• Affinity increased with an increase in chain length
– Similar changes in the enthalpic and entropic components.
Structural analysis
Orientation of the P1 and P2
parts of the molecules is
very similar
• Significant differences in
the interaction between
the terminal part of the P3
side chain and the binding
pocket.
• A combination of charge
repulsion, H-bonds, and
hydrophobic interactions
explain the observed
kinetic and
thermodynamic profiles
Conclusions
• Hydrogen bond formation and breakage are not necessarily
reflected in enthalpy gains and losses
• Changes in the structure of a lead compound can have
significant effects on its interaction with the target that
translate directly into kinetic and thermodynamic effects
Aiming for long residence time –
the blessing and curse of tight-binding inhibitors
Renin and MMP-12
Relevance of kinetics for clinical efficacy?
The renin-angiotensin cascade
Resolving the kinetics of renin inhibitors
Aliskiren
Ki = 0.04 nM
IC50 = 0.29 nM
O
HN
N
OH
H2N
O
O
O
H
N
NH2
N
H
S
O
O
O
H
N
OH
O
OH
O
aliskiren
Remikiren
remikiren
O
OH
H
Compound 1
Ki = 1.2 nM
IC50 = 14 nM
H
N
N
F
Ki < 0.04 nM
IC50 = 0.18 nM
HN
O
Cl
compound 1
F
Kinetics of tight binding renin inhibitors
Aliskiren displays long-lasting interactions with human renin
-1
-50
t (s)
20
20
Remikiren
Response (RU)
Response (RU)
Aliskiren
Response (RU)
14
-1
1000 -50
-1
1000 -50
t (s)
Compound 1
1000
t (s)
Compound
kon
(x105 M-1s-1)
koff
(x10-3 s-1)
Residence
time (min)
KD
(nM)
Ki
(nM)
IC50
(nM)
Compound 1
23 ± 2
4.9 ± 0.7
3.4
2.2 ± 0.3
1.2
14
Remikiren
7.4 ± 0.8
0.18 ± 0.05
93
0.24 ± 0.04
<0.04
0.18
Aliskiren
4.0 ± 0.2
0.11 ± 0.04
152
0.30 ± 0.09
0.04
0.29
Zinc-chelating hydroxamate MMP-12 inhibitors
Ca2
+
Ca2+
Zn2+
Ca2+
Zn2+
S
Batimastat
S
O
H
N
H
N
HO
Zinc binding
hydroxamate
group
N
H
O
O
Kinetics of hydroxamate MMP-12 inhibitors
Normalized response (RU)
45
35
5
4
25
15
Hydroxamates
3
5
-5
1
-15
-50
50
150
t (s)
250
350
Carboxylate
Comparison of hydroxamate and carboxylate inhibitors
15
Response (RU)
10
2
5
0
-5
1
-10
-15
-50
50
150
250
t (s)
350
450
Comparison of hydroxamate and carboxylate inhibitors
Conclusions
Keys to success
1. Careful experimental design
1.
2.
Sensor surface design, reference surfaces and compounds
Experimental conditions
2. Thorough data analysis
1.
2.
Raw data transformation into significant data
Definition of hit criteria (binding yes/no, vs. mechanistically
reasonable interaction characteristics)
3. The art of sample handling and appropriate controls
Implementation of SPR biosensor analysis for drug discovery
+ The technology is well established
+ The high information content is recognized
- Implementation is sometimes problematic
The user friendliness of commercial SPR biosensor instruments is
deceptive:
• Ease of use is not the same thing as ease of implementing the
technology for actual projects
• Challenges in all steps from experimental design to
interpretation of data
Acknowledgements
Beactica AB
• Thomas Gossas
• Per Källblad
• Gun Stenberg
Medivir AB
• Vera Baraznenok
• Ian Henderson
• Erik Lindström
Uppsala University
• Johan Winquist
• Helena Nordström
•
•
•
•
AstraZeneca
• Johanna Deinum
• Stefan Geschwindner
• Lars Gustavsson
• Djordje Musil
• Yafeng Xue
GE Healthcare/Biacore
• Markku Hämäläinen
Susanne Nyström
Lotta Vrang
Christer Sahlberg
Susanne Sedig
• Hans Wallberg
Chinese Academy of Sciences
• Ming-Hua Xu
• Zhi-Hua Sun
• Guo-Qiang Lin
Interactions understood. Leads improved.