Antitumor Activities and QSAR Analysis of
Antitumor Activities and QSAR Analysis of Piperazinediones
Piperazinediones
Hui
Hui--po Wang,po Wang,11Yu J.Tseng,Yu J.Tseng,22ChiunChiun--Li Wang,Li Wang,11TongTong--Ruen Wu,Ruen Wu,22JihJih--Hwa GuhHwa Guh33
Taipei Medical University College of Pharmacy,
Taipei Medical University College of Pharmacy,1 1 Graduate Institute of Biomedical Electronics and BioinformaticsGraduate Institute of Biomedical Electronics and Bioinformatics22and School of Pharmacy,and School of Pharmacy,33National Taiwan University,National Taiwan University, Taipei, TaiwanTaipei, Taiwan
I t
d
ti
I t
d
ti
R
R
lt
lt
d di
d di
i
i
ACS Annual Conference
ACS Annual Conference--San Francisco 20090301
San Francisco 20090301
Introduction
Introduction
Twenty four piperazinediones, screened on MDA Twenty four piperazinediones, screened on MDA--MBMB--231 231 breast cancer cell line, exhibited profound anticancer breast cancer cell line, exhibited profound anticancer activities with IC
activities with IC5050ranged between 10ranged between 10--66to 10to 10--88M (Teng and M (Teng and
Wang etal, US patent no. US Patent no. 6,635,649 B2, Fig. 1). Wang etal, US patent no. US Patent no. 6,635,649 B2, Fig. 1). The lead TW01 demonstrated (1) inhibition on 18 cancer The lead TW01 demonstrated (1) inhibition on 18 cancer--related serine/threonine kinases (IC
related serine/threonine kinases (IC50501.73 1.73 μμM/PKBa/Akt1, M/PKBa/Akt1,
Results and discussion
Results and discussion
phosphatase (IC
phosphatase (IC50501 ~ 0.1 1 ~ 0.1 μμM) with activity M) with activity
comparable to that of Gleevec (Table 1). A receptor comparable to that of Gleevec (Table 1). A receptor--independent 4D
independent 4D--QSAR paradigm was adopted for QSAR paradigm was adopted for constructing the structure
constructing the structure--based quantitative based quantitative inhibition models as a function of conformation, inhibition models as a function of conformation, alignment, and putative enzyme
alignment, and putative enzyme--binding, as well as binding, as well as
A four term model (Eq 1) and a five term model (Eq A four term model (Eq 1) and a five term model (Eq 2) were obtained. Reliability of the models are depicted 2) were obtained. Reliability of the models are depicted as the residual plot between predicted and observed as the residual plot between predicted and observed response, with the largest difference (
response, with the largest difference (––logIClogIC5050) < 0.23 ) < 0.23
in the case of BCM
in the case of BCM--TWTW--032032 (Fig. 4). The 3D (Fig. 4). The 3D pharmacophores suggested that non
pharmacophores suggested that non--polar type atom polar type atom is significant on the right arm of the analogues. is significant on the right arm of the analogues.
Possible pharmacophore binding models were Possible pharmacophore binding models were postulated (Fig. 5). Aromatic and hydrophobic postulated (Fig. 5). Aromatic and hydrophobic
pharmacohpores spreads over the binding site, similar to pharmacohpores spreads over the binding site, similar to the ATP binding proposed by McGregor on
the ATP binding proposed by McGregor on pharmacophore mapping of 220 inhibitors on protein pharmacophore mapping of 220 inhibitors on protein kinases. The piperazinediones might bind to the ATP kinases. The piperazinediones might bind to the ATP binding site of tyrosine kinases corresponding to the binding site of tyrosine kinases corresponding to the
50 50 μμ
2.32
2.32 μμM/Fyn, 1.4 M/Fyn, 1.4 μμM/Erk & 11.1 M/Erk & 11.1 μμM/MEK)M/MEK), , PI 3 kinasesPI 3 kinases, , abl abl tyrosine kinase and tyrosine
tyrosine kinase and tyrosine
TW -0 1 contr o l FB S Ta x o l 36789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 27 28 T W -01003 TW-01001 0 50 100 150 200 C e ll v ia b ilit y ( % o f c o n tr o l) Fi 1 A ti ti iti f 24 i i di Fi 1 A ti ti iti f 24 i i di
for searching the pharmacophore of this series of for searching the pharmacophore of this series of inhibitors.
inhibitors.
TW01
TW01 GleevecGleevec
Tyrosine Kinase
Tyrosine Kinase AblAbl 0.78 uM0.78 uM 0.25 uM*0.25 uM*
K562 CML
K562 CML 0.03 uM 0.03 uM 0.470.47uM**uM**
g g g
g g g
Eq 1:
Eq 1: --log IClog IC5050= 5.88 + 3.76*GCOD(3,= 5.88 + 3.76*GCOD(3,--4,4,--1,np) + 1,np) + --1.66*GCOD 1.66*GCOD (2,
(2,--3,0,np)+1.40* GCOD(1,6,2,any) + 3,0,np)+1.40* GCOD(1,6,2,any) + --1.19*GCOD(2,2,2,any) 1.19*GCOD(2,2,2,any) N=24 R
N=24 R2 2 = 0.897 Q= 0.897 Q22= 0.825 LOF = 0.024 = 0.825 LOF = 0.024 Eq 2:
Eq 2: --log IClog IC5050= 5.89 + 3.74*GCOD(3,= 5.89 + 3.74*GCOD(3,--4,4,--1,np) + 1,np) + --1.79*GCOD 1.79*GCOD (2,
(2,--3,0,np) +1.30*GCOD(1,6,2,np) + 3,0,np) +1.30*GCOD(1,6,2,np) + --1.18*GCOD(2,2,2,any)1.18*GCOD(2,2,2,any) +0.29*GCOD(2, +0.29*GCOD(2,--4,4,--1,np) 1,np) N=24 R N=24 R22= 0.906 Q= 0.906 Q22= 0.818 LOF = 0.028= 0.818 LOF = 0.028 Residual Plot g y p g g y p g “non
“non--type” GCODs discovered by Eq 1 and Eq 2. type” GCODs discovered by Eq 1 and Eq 2. Direction of structural modifications for activity are Direction of structural modifications for activity are proposed. The active conformation are much contributed proposed. The active conformation are much contributed by GCOD (
by GCOD (--2,1,10,np), implying that the non2,1,10,np), implying that the non--polar polar property for analogues is important especially on the property for analogues is important especially on the right arm. The model is used to predict the structures for right arm. The model is used to predict the structures for Optimization of the activity (Fig. 6).
Optimization of the activity (Fig. 6).
Table 1.
Table 1. Comparison of TW01 with Gleevec.Comparison of TW01 with Gleevec.
.*Buchdunger E, Biochim Biophys Acta, 2001. .*Buchdunger E, Biochim Biophys Acta, 2001.
(a) Fig. 1. Anticancer activities of 24 piperazinediones
Fig. 1. Anticancer activities of 24 piperazinediones against MDA
against MDA--MBMB--231 human231 human breast CA.breast CA.
Method
Method
Receptor
Receptor--independent 4Dindependent 4D--QSAR FormalismQSAR Formalism QSAR models are constructed based on IC QSAR models are constructed based on IC5050values values
(Fig. 2). The inhibitory measures are skewed toward potent (Fig. 2). The inhibitory measures are skewed toward potent inhibition. The range of
inhibition. The range of ––log IClog IC5050values of the training set values of the training set
are small and spans 2 orders of magnitude (7.54 to 5.3). are small and spans 2 orders of magnitude (7.54 to 5.3).
p
p--(polar (polar -- atoms); hba (hydrogen bond acceptor atoms); hba (hydrogen bond acceptor
atoms); hbd (hydrogen bond donor atoms); atoms); hbd (hydrogen bond donor atoms); ar(aromatic atoms). Sets of 13 trial alignments were ar(aromatic atoms). Sets of 13 trial alignments were used in constructing the 4D
used in constructing the 4D--QSAR models. The QSAR models. The selection criteria
selection criteria was with correlation coefficientscorrelation coefficients
Residual Plot 0 1 2 3 4 5 6 7 BCM -TW -024 BCM -TW-026 BCM -TW-027 BCM -TW-028 BCM -TW-029 BCM -TW-030 BCM -TW-031 BCM -TW-032 BCM -TW -03 3 BCM -TW -03 4 BCM -TW -035 BCM -TW-036 BCM -TW-037 BCM -TW-038 BCM -TW-039 BCM -TW-040 BCM -TW -041 BCM -TW -042 BCM -TW -04 3 BCM -TW -04 4 BCM -TW -045 BCM -TW-046 BCM -TW-049 BCM -TW-051 -l og I C 50 Observed Predicted Eq 4 Eq 5 any Non-polar Polar H-bond acceptor
Fig. 5. Predicted active Fig. 5. Predicted active
f i i ( )
f i i ( ) E 1E 1 d (b)d (b)
Fig. 6. (a) Predicted active (
Fig. 6. (a) Predicted active (––log log
IC
IC50507.19) and (b) predicted7.19) and (b) predicted
**
** Gottschalk S, Clin Can Res, 2004Gottschalk S, Clin Can Res, 2004.. (a)
(b)
p g ( )
p g ( )
GCOD (grid cell occupation descriptors) for building GCOD (grid cell occupation descriptors) for building quantitative models as a function of conformation, quantitative models as a function of conformation, alignment, and putative binding pharmacophore (IPE, Fig. alignment, and putative binding pharmacophore (IPE, Fig. 3). GCOD IPE types are represented as any
3). GCOD IPE types are represented as any((all atoms in (all atoms in the molecule); np (nonpolar atoms); p+ (polar
the molecule); np (nonpolar atoms); p+ (polar++atoms); atoms);
selection criteria
selection criteria was with correlation coefficients correlation coefficients R
R22>0.8 and the cross>0.8 and the cross--validation correlation validation correlation
coefficient Q
coefficient Q22> 0.7.> 0.7. Significant models should reflect Significant models should reflect
subtle differences in training set compounds leading subtle differences in training set compounds leading to differentiation in high activity.
to differentiation in high activity.
O N N N H O O O N N H N H O O O N N H N H O O O OH N N H N H O O O O BCM-TW-024 BCM-TW-026 BCM-TW-027 BCM-TW-028
IC50=14% IC50=82.6% IC50=60.5% IC50=44.2%
-BCM-TW-037 BCM-TW-038 BCM-TW-039 BCM-TW-040 BCM-TW-033 BCM-TW-034 BCM-TW-035 BCM-TW-036 N N H N H O O O F N N H N H O O O Cl N N H N H O O O O N N H N H O O O O BCM-TW-029 BCM-TW-030 BCM-TW-031 BCM-TW-032
IC50=79.7% IC50=46.5% IC50=76.7% IC50=18.6%
N N H N H O O O a b c N N H N H O O O a b c N N H N H O O O a b c N N H N H O O O a b c 1 2 3 4 N H O O ca N H O O c a N H O O b a N H O O B B B B B B B B B B B B B B B B B B B B B B B B Compound Name
Fig. 4. A plot of the difference (residual) in predicted Fig. 4. A plot of the difference (residual) in predicted activity (Eq 1) and the observed activity of TW01 activity (Eq 1) and the observed activity of TW01 analogues analyzed.
analogues analyzed.
conformation using (a)
conformation using (a) Eq 1 and (b) Eq 1 and (b) Eq 2
Eq 2 with alignment 3. Green grid with alignment 3. Green grid cells are IPE type “any” and white cells are IPE type “any” and white grid cells are IPE type “non
grid cells are IPE type “non--polar”.polar”.
IC
IC50 50 7.19) and (b) predicted 7.19) and (b) predicted inactive (
inactive (––log IClog IC50504.10) using 4.10) using Eq 2. Blue grid cells are those Eq 2. Blue grid cells are those contribute positively to inhibition contribute positively to inhibition potency, while the red grid cells potency, while the red grid cells reduce potency.
reduce potency.
Conclusion
Conclusion
References
References
The piperazinediones are assumed to bind to The piperazinediones are assumed to bind to protein kinases based on the inhibition profile of the protein kinases based on the inhibition profile of the
1. Correspondence:
1. Correspondence: HuiHui--po Wang (po Wang (hpw@tmu.edu.tw)hpw@tmu.edu.tw)
2 Hopfinger A J ; Tseng Y ; etal Evaluation of Alignment 2 Hopfinger A J ; Tseng Y ; etal Evaluation of Alignment
N N H N H O O O N+O O N N H N H O O O Cl N N H N H O O O O O N N H N H O O O Cl Cl
IC50=62.2% IC50=105.8% IC50=132.6% IC50=58.1%
N NH N H O O O O O N N H NH O O O N N H NH O O O OH N N H NH O O O O BCM-TW-041 BCM-TW-042 BCM-TW-043 BCM-TW-044
IC50=109.3% IC50=29.1% IC50=4.7% IC50=22.1%
N N H N H O O O F N N H N H O O O Cl N N H N H O O O S N N H N H O O O NH O BCM-TW-045 BCM-TW-046 BCM-TW-049 BCM-TW-051
IC50=27.9% IC50=33.7% IC50=50.0% IC50=53.5%
N N H N H O O O S N N H N H O O O NH O N N H N H O O O N+ O -O
IC50=120.9% IC50=107.0% IC50=56.4% IC50=54.7%
O N H H N O O N N N H O O O b N N N H O O O b N N N H O O O b c N N N H O O O b a c 5-1 5-2 6-1 6-2 7-1 7-2 8 N N H N H O O Oc b a 9 N N H N H O O O a b c 10
protein kinases based on the inhibition profile of the protein kinases based on the inhibition profile of the lead TW01. Our receptor
lead TW01. Our receptor--independent 4Dindependent 4D--QSAR QSAR analysis indicated that the non
analysis indicated that the non--polar property of polar property of piperazinediones
piperazinediones has important contribution for has important contribution for binding
binding to the ATP site of protein kinasesto the ATP site of protein kinases according to according to McGregor’s pharmacophore mapping of 220 protein McGregor’s pharmacophore mapping of 220 protein kinase inhibitors,
kinase inhibitors,. .
2. Hopfinger A. J.; Tseng, Y.; etal, Evaluation of Alignment 2. Hopfinger A. J.; Tseng, Y.; etal, Evaluation of Alignment
Dependence in 3D
Dependence in 3D--QSAR Model Construction Using QSAR Model Construction Using 4D
4D--QSAR Analysis”, Internet J Chem, 2000, 3, 10.; QSAR Analysis”, Internet J Chem, 2000, 3, 10.; 3. McGregor, M. J., A Pharmacophore Map of Small 3. McGregor, M. J., A Pharmacophore Map of Small
Molecule Protein Kinase Inhibitors, J. Chem. Inf. Model, Molecule Protein Kinase Inhibitors, J. Chem. Inf. Model, 2007, 47(6), 2374
2007, 47(6), 2374--2382.2382.
Fig. 2. IC50against MDAMDA--MBMB--231 breast cancer cell line231 breast cancer cell line
Fig. 3. Set of 13 trial alignments used in constructing Fig. 3. Set of 13 trial alignments used in constructing 4D