Adventures in Computational Enzymology

1
Adventures in Computational Enzymology

• John Mitchell
• University of St Andrews

2
The MACiE Database
Mechanism, Annotation and Classification in
Enzymes. http//www.ebi.ac.uk/thornton-srv/databas
es/MACiE/
Gemma Holliday, Daniel Almonacid, Noel OBoyle,
Janet Thornton, Peter Murray-Rust, Gail
Bartlett, James Torrance, John Mitchell
G.L. Holliday et al., Nucl. Acids Res., 35,
D515-D520 (2007)
3
Enzyme Nomenclature and Classification
4
The EC Classification

• Deals with overall reaction, not mechanism

• Reaction direction arbitrary
• Cofactors and active site residues ignored
• Doesnt deal with structural and sequence
  information
• However, it was never intended to do so

5
A New Representation of Enzyme Reactions?

• Should be complementary to, but distinct from,
  the EC system
• Should take into account
• Reaction Mechanism
• Structure
• Sequence
• Active Site residues
• Cofactors

• Need a database of enzyme mechanisms

6
MACiE Database
Mechanism, Annotation and Classification in
Enzymes. http//www.ebi.ac.uk/thornton-srv/databas
es/MACiE/
7
(No Transcript)
8
(No Transcript)
9
Global Usage of MACiE
10
MACiE Entries
11
MACiE Mechanisms are Sourced from the Literature
12
Coverage of MACiE
Representative based on a non-homologous
dataset, and chosen to represent each available
EC sub-subclass.
13
EC is not Everything

• Different mechanisms can occur with exactly the
  same EC number.
• MACiE has six beta-lactamases, all with different
  mechanisms but the same overall reaction.

14
(No Transcript)
15
EC Coverage of MACiE
Structures exist for 6 EC 1.-.-.- 61 EC
1.2.-.- 204 EC 1.2.3.- 1776 EC 1.2.3.4
MACiE covers 6 EC 1.-.-.- 57 EC 1.2.-.-
183 EC 1.2.3.- 321 EC 1.2.3.4
Representative based on a non-homologous
dataset, and chosen to represent each available
EC sub-subclass.
16
EC Coverage of MACiE
17
Repertoire of Enzyme Catalysis
G.L. Holliday et al., J. Molec. Biol., 372,
1261-1277 (2007) G.L. Holliday et al., J. Molec.
Biol., 390, 560-577 (2009)
18
(No Transcript)
19
(No Transcript)
20
Repertoire of Enzyme Catalysis
Enzyme chemistry is largely nucleophilic
21
Repertoire of Enzyme Catalysis
Enzyme chemistry is largely nucleophilic
22
Repertoire of Enzyme Catalysis
23
Repertoire of Enzyme Catalysis
24
Repertoire of Enzyme Catalysis
25
Repertoire of Enzyme Catalysis
26
Repertoire of Enzyme Catalysis
27
Repertoire of Enzyme Catalysis
We do see a few steps corresponding to well-known
organic reactions but these are the exception.
28
Residue Catalytic Propensities
29
Residue Catalytic Functions
30
Phospholipidosis
Lowe et al., Molec. Pharmaceutics, 7, 1708 (2010)

• An adverse effect caused by drugs
• Excess accumulation of phospholipids
• Often by cationic amphiphilic drugs
• Affects many cell types
• Causes delay in the drug development process

31
Phospholipidosis
Lowe et al., Molec. Pharmaceutics, 7, 1708 (2010)

• Causes delay in the drug development process
• May or may not be related to human pathologies
  such as Niemann-Pick disease

32
Electron micrographs of alveolar macrophages (A
and B) and peritoneal macrophages (C and D)
obtained from 3-month-old Lpla2/ and Lpla2-/-
mice
Hiraoka, M. et al. 2006. Mol. Cell. Biol.
26(16)6139-6148
33
Tomizawa et al.,
34
Literature Mined Dataset

• Produced our own dataset of 185 compounds (from
  literature survey)
• 102 PPL and 83PPL-
• Each compound is an experimentally confirmed
  positive or negative

R. Lowe, R.C. Glen, J.B.O. Mitchell Mol. Pharm.
2010 VOL. 7, NO. 5, 17081714
35
Some PPL molecules, from Reasor et al., Exp Biol
Med, 226, 825 (2001)
36
10001101010011001101
10110101000011101101
10111101010001001100
10000001110011100111
10100101011101001110
10011111110001001010
Represent molecules using descriptors (we used
E-Dragon Circular Fingerprints)
37
Experimental Design
Split data into N folds, then train on (N-2) of
them, keeping one for parameter optimisation and
one for unseen testing. Average results over all
runs (each molecule is predicted once per N-fold
validation). We also repeat the whole process
several times with randomly different assignments
of which molecules are in which folds.
38
Models are built using machine learning
techniques such as Random Forest
39
or Support Vector Machine
40
(No Transcript)
41
Results
Average MCC Values RF SVM 0.619 0.650
42
So we have built a good predictive model that can
learn the features that predispose a molecule to
being PPL, and can make predictions from
chemical structure. This is useful one could
add it to a virtual screening protocol. But can
we understand anything new about how
phospholipidosis occurs?
43
Read up on gene expression studies related to
phospholipidosis
44
Sawada et al. listed genes which they found to be
up- or down- regulated in phospholipidosis
45
As with all gene expression experiments, some of
these will be highly relevant, others will be
noise. Can we help interpret these data?
46
Mechanism?
H. Sawada, K. Takami, S. Asahi Toxicological
Sciences 2005 282-292
47

• What expertise do we have available amongst our
  team, colleagues collaborators?
• Multiple target prediction
• Maths
• Programming

Florian Nigsch
Hamse Mussa
Rob Lowe
48

• Multiple target prediction
• Predicting off-target interactions of drugs. Not
  with the primary pharmaceutical target, but with
  other targets relevant to side effects.

49
CHEMBL
Data mining and filtering
Filtered CHEMBL, 241145 compounds 1923 targets
Random 991 split of the whole dataset, 10 repeats
10 models
Phospholipidosis dataset 100 PPL, 82 PPL-
compounds
Predicted target associations
Target PS? scores
50
ChEMBL Mining

• Mined the ChEMBL (03) database for compounds and
  targets they interact with
• Target description included the word "enzyme",
  "cytosolic", "receptor", "agonist" or "ion
  channel"
• A high cut-off (weak binding) was used on
  Ki/Kd/IC50 values (lt 500µM) to define activity

51
Method

• Number of Compounds 241145
• Number of Targets 1923
• Split the data into 10 different partitions of
  training and validation
• Used circular fingerprints with SYBYL atom types
  to define similarities between molecules

52
Multi-class Classification

• Algorithms
• Parzen-Rosenblatt window
• Naive Bayes

53
Parzen-Rosenblatt window

• Rank likely targets using estimates of
  class-condition probabilities

using a Gaussian kernel K(xi, xj)
(xi - xj)T(xi - xj) corresponds to the number of
features in which xi and xj disagree
54
Partition No. PRW Rank NB Rank
1 17.049 74.104
2 16.343 76.251
3 18.424 79.078
4 16.212 73.539
5 17.339 73.535
6 18.630 77.244
7 20.694 78.560
8 18.870 74.464
9 16.584 76.235
10 18.200 78.077
Average 17.835 76.109
When we test the two methods, PRW ranks known
targets better than Naïve Bayes does. Hence we
use PRW for our study.
55
Assemble List of Targets Relevant to Sawadas
Suggested Mechanisms
Mechanisms 1. Inhibition of lysosomal
phospholipase activity 2. Inhibition of
lysosomal enzyme transport 3. Enhanced
phospholipid biosynthesis 4. Enhanced
cholesterol biosynthesis.
56
Assemble List of Targets Relevant to Sawadas
Suggested Mechanisms
Inhibition of lysosomal phospholipase activity
Enhanced phospholipid biosynthesis
Enhanced cholesterol biosynthesis
57
Assigning Scores to Targets

• Use these 10 models of target interactions
• Predict targets for phospholipidosis dataset
• Score targets according to the likelihood of
  involvement in phospholipidosis
• Use the top 100 predicted targets per compound as
  we seek off-target interactions

58

• Score measures tendency of target to interact
  with PPL rather than PPL- compounds.

59
(No Transcript)
60
M1 M5 are involved in phospholipase C
regulation may be relevant but not in Sawadas
list.
61
(No Transcript)
62
62
We consider a PS? score significant if the target
is predicted to interact with at least 50 more
PPL compounds than PPL- compounds.
63
Our Scores for 8 of Sawadas PPL-Relevant Targets
Mechanism Target Rank PS?
1 Sphingomyelin phosphodiesterase (SMPD) (h) 225 55
  Lysosomal Phospholipase A1 (LYPLA1) (r) 163 90
  Phospholipase A2 (PLA2) (h) 152 97
3 Elongation of very long chain fatty acids protein 6 (ELOVL6) (h) 1203 -10
  Acyl-CoA desaturase (SCD) (m) 610 0
4 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) (h) 456 10
  Squalene monooxygenase (SQLE) (h) 437 14
  Lanosterol synthase (LSS) (h) 114 134
Inhibition of lysosomal phospholipase activity
Enhanced phospholipid biosynthesis
Enhanced cholesterol biosynthesis
64
Our Scores for Sawadas PPL-Relevant Targets
Mechanism Target Rank PS?
1 Sphingomyelin phosphodiesterase (SMPD) (h) 225 55
  Lysosomal Phospholipase A1 (LYPLA1) (r) 163 90
  Phospholipase A2 (PLA2) (h) 152 97
3 Elongation of very long chain fatty acids protein 6 (ELOVL6) (h) 1203 -10
  Acyl-CoA desaturase (SCD) (m) 610 0
4 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) (h) 456 10
  Squalene monooxygenase (SQLE) (h) 437 14
  Lanosterol synthase (LSS) (h) 114 134
Inhibition of lysosomal phospholipase activity
Enhanced phospholipid biosynthesis
Enhanced cholesterol biosynthesis
65
Other Mechanisms

• The mechanisms and targets suggested here are
  insufficient to explain all the PPL compounds in
  our data set.
• We expect that other targets and possibly
  mechanisms are important.
• Our method cant test direct compound
  phospholipid binding.

66
(No Transcript)
67
67
68
ACKNOWLEDGEMENTS
Dr Gemma Holliday Dr Rob Lowe Dr Daniel
Almonacid Prof. Janet Thornton Dr Florian
Nigsch Dr Hamse Mussa Prof. Bobby Glen Dr Andreas
Bender Alexios Koutsoukas
69
ACKNOWLEDGEMENTS