“Quantitative Structure Activity Relationship (QSAR)...

65
A pan-European Infrastructure for Quality in Nanomaterials Safety Testing NUID UCD | NHM | IOM | JRC | BFR | KIT | FUNDP | IST | UNIVLEEDS | NILU | HMGU | LMU | CIC | UU | ICN DLO/RIKILT | WU | DGUV | TAU | VITO | SMU | TCD | FIOH | UOE | HW | CNRS | INERIS | UBRM | RIVM “Quantitative StructureActivity Relationship (QSAR) models” 27.03.2013 Xue Z. Wang Ceyda Oksel University of Leeds

Transcript of “Quantitative Structure Activity Relationship (QSAR)...

Page 1: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

A pan-European Infrastructure for Quality in Nanomaterials Safety Testing

NUID UCD | NHM | IOM | JRC | BFR | KIT | FUNDP | IST | UNIVLEEDS | NILU | HMGU | LMU | CIC | UU | ICN DLO/RIKILT | WU | DGUV | TAU | VITO | SMU | TCD | FIOH | UOE | HW | CNRS | INERIS | UBRM | RIVM

“Quantitative Structure‐Activity Relationship (QSAR) models”

27.03.2013

Xue Z. WangCeyda OkselUniversity of Leeds

Page 2: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

• Theory of QSAR• New Data Mining Methods• Nanomaterial Characterization• Nanotoxicity• Application of QSAR model in nanotoxicity modeling

2QualityNano Training School 

Purpose of Today

Page 3: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Part ITheory of QSAR 

3QualityNano Training School 

Page 4: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

QSAR History

Corwin Herman Hansch(1918 – 2011)

The father of the concept of quantitative structure‐activity relationship (QSAR),the quantitative correlation of the physicochemical 

properties of molecules with their biological activities[5]

4

1865

Crum Brown andFraser[1] 

expressed theidea that there

was a relationship between activityand chemicalstructure.

1900

Meyer[3] and Overton[4] found 

linear relationshipsbetween the 

toxicity of organic compounds and their lipophilicity.

1969

Hansch[5]  published a free‐energy relatedmodel to correlate biological activities 

withphysicochemical

Properties.

1893

Richet[2] correlated toxicities of 

simple organicmoleculeswith their solubility in 

water.

[1] A. Crum‐Brown, T.R. Fraser, Trans. R. Soc. Edinb. 25 (1968–1969) 257. [2] M.C. Richet, Compt. Rend. Soc. Biol. 45 (1893) 775.[3] H. Meyer, Arch. Exp. Pathol. Pharmakol. 42 (1899) 109. [4] E. Overton, Studien u¨ber dir Narkose, Gustav Fischer, Jena, 1901[5] Hansch, C. (1969) A quantitative approach to biochemical structure‐activity relationships Accounts of Chemical Research, 2, 232‐239.

QualityNano Training School 

Page 5: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Molecular Structure

Physicochemicalor Biological Property

The Variance In

Mathematical Dependencies

Toxicity/Ecotoxicity Prediction

Bulk Form                 Nano‐particles(widely used)                        (not adopted)

A QSAR is a statistical modelthat relates a set of structuraldescriptors of a chemicalcompound to its biologicalactivity.

5

QSAR (Quantitative Structure‐Activity Relationship)

What is QSAR ?

[1] T. Puzyn et al. (eds.), Recent Advances in QSAR Studies, 103–125. DOI 10.1007/978‐1‐4020‐9783‐6_4, C  Springer Science+Business Media B.V. 2010QualityNano Training School 

The presence of particular characteristics increasethe propability that the compund is toxic

Page 6: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

QSAR (Quantitative Structure‐Activity Relationship)nano‐QSAR

QSAR models are very useful in case of the classic chemicals but theconcept of nano‐QSAR is still under development.

QSAR has been widely used to predict the toxicity of substances in bulkform (especially drug‐like compunds) but, up to date, QSAR studies for theprediction of nanoparticle toxicity have been rarely reported.

Successful QSAR studies What about Nano‐size particles?

6QualityNano Training School 

The European system for new chemical management (REACH) promotesQSAR methods as an alternative way of toxicity testing.

Page 7: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

IN SILICO IN VITRO                            IN VIVO(computational) (in container) (on living organism)

predict predict

All chemical substances need to be tested in terms of their toxicologicaland environmental properties before their useThere are several reasons to use QSAR Models : very fast, often free, reduce the number of animals used in experiments 

validate

The most important sources of information

Alternative, fastdeveloping applications

Reduction in the time, cost and animal testing

the mostpromising

QSARquantitative structure‐activity 

relationship[1]

7[1] A.. Gajewicz, et al., Advancing risk assessment of engineered nanomaterials: Application of computational approaches,Adv. Drug Deliv. Rev. (2012), doi:10.1016/j.addr.2012.05.014[1]

Sources of information[1]

Why do we need QSAR models?

QualityNano Training School 

Page 8: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

8

Purpose of QSARTo predict biological activity and physico‐chemical properties by rational means

To comprehend and rationalize the mechanisms of action within a series of chemicals

Savings in the cost of product development

Predictions could reduce the requirement for lengthy and expensive animal tests.

Reduction of animal tests, thus reducing animal use and obviously pain and discomfort to animals

Other areas of promoting green and greener chemistry toincrease efficiency and eliminate waste

QualityNano Training School 

Page 9: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

9

Requirements for QSAR

How many compounds are required to develop a QSAR?

There is no direct and simple answer

“as many as possible”

To provide some guide, it is widely accepted that between five and ten compounds are required for every descriptor in a QSAR [1]

[1] Aptula AO, Roberts DW(2006) Mechanistic applicability domains for non‐animal based prediction of toxicological end points: General principles and application to reactive toxicity. Chem Res Tox. 19:1097–1105

1.Set of  2. Set of molecular 3. Measured biological 4.Statisticalmolecules descriptors activity or property methods

Toxicity EndpointsMolecular

Modellinge.g. Neural NetworksPLS, Expert Systems

DESCRIPTORSPhyscochemical, biological, structural

QSARs

Daphnia magna EC50

CancinogenicityMutagenicity

Rat oral LD50Mouse inhalation

LC50Skin sensitisation

Eye irritancy

QualityNano Training School 

Page 10: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Applications of QSAR

10

There are a large number of applications of these models withinindustry, academia and governmental (regulatory) agencies.

The estimation of physico‐chemical properties, biologicalactivities and understanding the physicochemical featuresbehind a biological response in drug desing.

The rational design of numerous other products such as surface‐active agents, perfumes, dyes, and fine chemicals.

The identification of hazardous compounds at early stages of product development, the prediction of toxicity to humans and environment.

The prediction of fate of molecules which are releasedinto the environment.

The prediction of a variety of physico‐chemical properties of molecules.

QualityNano Training School 

Page 11: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

11

What is required for a good QSAR Model ?

The QSAR model should meet therequirements of the OECD principles [1]:

[1] OECD, Guidance Document on the Validation of (Quantitative) Structure– Activity Relationships Models, Organisation for Economic Co‐operation and Development (2007).

1 • a defined endpoint;

2 • an unambiguous algorithm;

3 • a defined domain of applicability ;

4• appropriate measures of goodness‐of‐fit, robustness and predictivity;

5 • a mechanistic interpretation if possible.

•Uninformative descriptors•Poor descriptor selection and chance correlations•Modelling complex, nonlinear structure propertyrelationships with linear models•Incorrectly validating QSPR models•Not understanding the domain of applicability of models

Common QSAR ModellingErrors

QualityNano Training School 

Page 12: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

12

Methods

Molecular Structure

Representation

Molecular Descriptors

Predictive Model

Model Validation

Applicability Domain of QSAR

Types of Molecular Descriptors(topological, geometric, electronic and hybrid)

1D Descriptors 2D Descriptors 3D Descriptors

Main Steps

•QSAR Modelling process consists of 5 main steps.

QualityNano Training School 

•Begins with the selection of molecules to be used•Selection of descriptors; numericalrepresenter of molecular features (e.g. numberof carbon)•Original descriptor pool must be reduced in size•Model building•The reliability of the model should be tested

Page 13: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

13

Molecular Descriptors[1]are numerical values that characterize properties of molecules.

cacalculated by

Mathematicalformulas

Computationalalgortihms

TheoreticalIndexes

Experimentalmeasurements

MolecularDescriptors

MolecularDescriptors

Different Theories(e.g. Quantum‐chemistry,     

information theory, organic chemistry)

derivedfrom Scientific Fields (e.g. QSAR,

medicinal chemistry, drug design, toxicology, analytical 

chemistry,Environmetrics, virtual 

screening)

appliedin

statisticschemometrics

chemoinformatics

processedby

[1] T. Puzyn et al. (eds.), Recent Advances in QSAR Studies, 103–125. DOI 10.1007/978‐1‐4020‐9783‐6_4, C  Springer Science+Business Media B.V. 2010[2] Todeschini R, Consonni V (2000) Handbook of molecular descriptors. Wiley‐VCH, Weinheim

More than 5000[2]

QualityNano Training School 

Page 14: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Regression Problems• Multiple linear regression• Partial least squares• Feed‐forward back propagation neural network• General regression neural networkClassification Problems• Linear Discriminant analysis• Logistic regression• Decision tree• K‐nearest neighbor• Probabilistic Neural Network• Support vector machineRecently‐developed Methods• Kernel partial least square• Robust continuum regression• Local lazy regression• Fuzzy interval number k‐nearest neighbor• Fast projection plane classifier

14

Data Analysis

Model Validation

Data Collection

MolecularDescriptor

[1] Ekins, S. (2007). Computational Toxicology: Risk Assessment for Pharmaceutical and Environmental Chemicals. New Jersey: Wiley-Interscience.

Data Analysis Methods[1]

QualityNano Training School 

Different Statistical Methods have been used in QSAR for the extraction of useful informationfrom the data.

Page 15: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Data Mining Methods•a relatively new field which share some techniques with traditional 

data analysis but also brings a lot of new visions and techniques. 

Data: records of numerical data, symbols, images,documents

Knowledge: Rules: IF .. THEN ..Cause‐effect relationshipsDecision treesPatterns: abnormal, normal     operationPredictive equations

Data Data Data Data

Information

Knowledge

Decision

Volume

Value

‘DM is the discovery of useful Informationand Knowledge from a large collection ofData, in order to help the Decisionmaking’

15QualityNano Training School 

Page 16: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

16

Clustering

Classification

Conceptual Clustering

Inductive learning

•Data mining shares many techniques and tools with traditional dataanalysis (such as the ones listed below). But it emphasises the analysisof very large databases, and can deal with complicated cases whichcannot be handled by traditional methods.

•Data mining research has also generated some new techniques thatwere not seen in traditional data analysis, such as Link Analysis

QualityNano Training School 

Summarisation

Regression

Case-based Learning

Data Mining Methods

Page 17: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

17

x1 x2 x31 0 01 1 10 0 11 1 10 0 00 1 11 1 10 0 01 1 10 0 0

x1 x2 x3

x1x2

x3

• Example : Ten data cases, three variables, x1, x2, and x3• How can we automatically uncover the dependency relationship between them ?

This relationship means, x2 isdependent on x1, and x3 is dependent on x2, but x1 is independent of x2 and x3

This relationship means, x2 isdependent on x1,  x3 is also dependent on x1, but x2 and x3 are independent of each other

DATA SET WHICH RELATIONSHIP ?

QualityNano Training School 

New DM Technique Example 1:  Link Analysis 

Page 18: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

18

x1 x2 x3 x4 y x1 x2 x3 x4 y55.33 1.72 54 1.66219 92.19 71.31 3.44 55 1.60325 90.5159.13 1.2 53 1.58399 92.74 72.3 4.02 55 1.66783 90.2457.39 1.42 55 1.61731 91.88 68.81 6.88 55 1.69836 91.0156.43 1.78 55 1.66228 92.8 66.61 2.31 52 1.77967 91.955.98 1.58 54 1.63195 92.56 63.66 2.99 52 1.81271 91.9256.16 2.12 56 1.68034 92.61 63.85 0.24 50 1.81485 92.1654.85 1.17 54 1.58206 92.33 67.25 0 53 1.72526 91.3652.83 1.5 58 1.54998 92.22 67.19 0 52 1.86782 92.16

x1, x2, x3 are different measures of feed compositionx4  is the log of a combination of operating conditionsy – product quality measure   y = ‘good’ if  > 92,  = ‘low’ if between 91‐92,= ‘very low’ if < 91

QualityNano Training School 

New DM Technique Example 2: Inductive Data Mining for Automatic Generation of Decision Trees from Data

Aim: to find which feed composition 

or operational condition variables 

have the most important impact on 

product quality

Page 19: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

19

If x1 = [44.6, 55.4], then conditional probability of y = 3 is 0.9

If x4 = [1.8, 2.3], then the conditional probability of y =3 is 1.0

x1

y > 92 x4

x1 Y > 92

91<y<92 Y < 91

Good

Good

Low Very low

1.5

2.0

50 60 70 x1

x4Very low Low Good

< 55.4 > 55.4

>1.8<1.8

< 70> 70

Product quality is mainly determined by x1 and x4, almost has nothing to do with x2 and x3•If x1 < 55.4, the probability that product quality is good (y>92) is high•If  x1 >55.4 but x4  >1.8,  the probability that product quality is good is also high•Other two regions, product quality is either low or very low

QualityNano Training School 

Automatically generated decision tree

Page 20: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

20

(1)Generation of a population of solutions

(2) Repeat steps (i) and (ii) until the stop criteria are satisfied:(i) calculate the fitness function values for each solution candidate

(ii) perform crossover and mutation to generate the next generation

(3) the best solution in all generations: the final solution

<s >s

.. .... ..

<s >s

.. .... ..

<s >s

.... ..

<s >s

.. .... ..… …

<s >s

.. .... ..

<s >s

.. .... ....

<s >s

.... ..

<s >s

.. .... .. … …

Automatic generation of 

decision tree from data

This is a newapprooach

based on a genetic programming 

method

QualityNano Training School 

Page 21: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

21

Data Set 1:Concentration lethal to 50% of the population, LC50, 1/Log(LC50), of vibrio fischeri, a biolumininescent bactorium

75 compounds 1069 molecular descriptors calculated by molecular modelling software DRAGON

Data Set 2:Concentration effecting 50% of the population, EC50 of algae chlorella vulgaris, by causing fluoresceindiacetate to disappear

80 compounds 1150 descriptors

Two Data Sets

QualityNano Training School 

New DM Technique Example

Page 22: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

22

Sample tree from Y‐Adapt

Class 1 (21/21)

Class 1 (8/8)

No

NoYes

Yes

Geary autocorrelation lag 4 weighted by atomic mass ≤ 

0.007

Polar surface area ≤ 40.368

Class 2 (9/9)

NoYes

H autocorrelation lag 0 weighted by Sanderson electro‐negativities ≤ 2.111

Geary autocorrelation lag 4 weighted by polarizability ≤ 

1.181

Class 3 (9/10)

Yes No

Broto‐Moreau autocorrelation lag 2 weighted by polarizability ≤ 

4.085

Class 2 (5/6)

Yes No

Class 3 (6/6)

A tree for the bacteria data using three 

classes.

For bacteria data:of 1069 descriptors, only 5 descriptors are important ! Using the 5 descriptors, the tree 

model can predict withthe accuracy >96.7%

QualityNano Training School 

Y‐Adapt for QSAR Toxicity Prediction: Data Set 1A method to classify toxicity endpoints into classes

Page 23: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

23

A decision tree from generation 46 for the bacteria data, finding fourclasses from the continuous endpoint itself during induction.

Training accuracy: 93.3%Test accuracy: 80.0%

NoYes

Yes

NoYes

Yes No

No NoYes

Class 1 (13/13)

3D MoRSE signal 15 weighted 18 weighted by Sanderson electronegativites ≤ ‐0.644

Geary autocorrelation lag 4 weighted by atomic mass ≤ 0.007

Polar surface area ≤ 40.368

H autocorrelation lag 0 weighted by Sanderson electro‐negativities ≤ 2.111

Class 2 (5/7)

Class 3 (7/9)

Class 4 (6/6)

Class 1 (16/16)

Class 2 (9/9)

H autocorrelation lag 0 weighted by Sanderson electro‐negativities ≤ 2.394

Class 1: Log(1/EC50) ≤  4.08       29 casesClass 2: 4.08 < Log(1/EC50) ≤ 4.51        16 casesClass 3: 4.51 < Log(1/EC50) ≤  4.99       9 casesClass 4: otherwise                        6 cases

QualityNano Training School 

Y‐Adapt for QSAR Toxicity Prediction : Data Set 1

Page 24: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

24

NoYes

Class 2(14/15)

Class 1(16/16)

No Yes

Class 3(6/8)

NoYes

Yes No

Self‐returning walk count order 8 ≤ 3.798

H autocorrelation of lag 2 weighted by Sanderson electro‐negativities ≤ 

0.401

Molecular multiple path count order 3 ≤ 92.813

Solvation connectivity index ≤ 2.949

Class 4(6/7)

NoYes

2nd component symmetry directional WHIM index

weighted by van der Waals volume ≤ 0.367

Class 3(9/10)

Class 4(8/8)

2nd dataset ‐ algae data

Of the 1150 descriptors, only 5 descriptors are important !

Training: 92.2%Test: 81.3% 

QualityNano Training School 

Y‐Adapt for QSAR Toxicity Prediction : Data Set 2

Page 25: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

25

Results PresentationGraphsTablesASCII files

Discovery ValidationStatistical significanceResults for training and test sets

Data Mining ToolboxRegression PCA and ICAART2 networksKohonen networksK‐nearest neighbourFuzzy c‐meansDecision trees and rulesFeedforward neural networks (FFNN)Summary statisticsVisualisationMixture QSAR models 

Data importExcelASCII FilesDatabaseXML

Data Pre‐processingScalingMissing valuesOutlier identificationFeature extraction

Descriptor calculation

QualityNano Training School 

Different data mining techniques can be 

combined into a single system

Integrated Process Data Mining Environment

Page 26: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

The statistical fit of model R2

Standart Error of Predicton (S)

Cross‐validation (Q2)• K‐fold Cros Validation• Leave‐one‐out

Cross Validation

26

ModelValidation

QSAR Model Validation

Model Applicability •Model validation is required to ensure 

model reliability (quality of the model)Ensures that the model is not due to 

chance factors!Model

Interpretation

•Cross validation: dividing a data sample into subsets, performing the analysis on a single subset, using other subsets to confirm and validate the initial analysis

QualityNano Training School 

•There is no agreed method for the validation of QSAR Models

•The statistical fit of model is usually performed

Page 27: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

27

QSAR Workflow[1]

[1] Tropsha, A., Gramatica, P., Gombar, V. (2003) The importance of being earnest:Validation is the Absolute Essential for Successful Application and Interpretation of QSPR Models. Quant. Struct. Act. Relat. Comb. Sci. 22, 69‐77. QualityNano Training School 

Page 28: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

28

QSAR Model Applicability Domain

.  .    . .

..  .    .    ... ..

Descriptor 1

...

..

.. ..

..TRAINING SETDescriptor 2

Inside ApplicabilityDomain

will be predicted

Outside ApplicabilityDomain

will not be predicted

•Not even a significant and validated QSAR model can be used for the entire universe of 

chemicals[1]!

[1] Gramatica P.  (2007) Principles of QSAR models validation: internal and external. QSAR and Combinatorial Science; 26:694–701.

•Model applicability allows us to decide whether the model will be useful for new data (correct use of 

the model).

•Finding the boundaries of themodel to see how well it will work

for other compounds.

QualityNano Training School 

Page 29: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

29

QSAR Tools for Toxicity Prediction1. TOPKAT (by Komputer Assisted Technology ):

It uses QSARs to give numeric predictions, based on several descriptors

2. HazardExpert (by CompuDrug ):HazardExpert, produced by CompuDrug, is a rule‐based system. The rules use toxicophores, or structural alerts derived from QSARs

3. CASE: It can derive QSAR model from user’sdata.

4. ECOSAR: It uses QSARs to predict the aquatic toxicity of chemicals for a variety of organisms

5. OASIS: It uses hierarchical rules based on QSARs, which can be user derived. 

6.     DEREK:Deductive Estimation of Risk from Existing Knowledge (DEREK) is marketed by LHASA Ltd. It is an expert knowledge base system that predicts whether a chemical is toxic

QualityNano Training School 

• Each of these QSAR tools have their own advantages and weaknesses for toxicitypredictions

• There are also several available software packages for descriptor calculations (e.g. DRAGON)

Page 30: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

30

Challenges for QSAR [1]

[1] Puzyn, T., Leszczynski, J.(2012) Towards Efficient Designing of Safe Nanomaterials: Innovative Merge of Computational Approaches and Experimental Techniques, The Royal Society of Chemistry. 

1

• If there is a measurement error in the experimental data, it is very likelythat false correlations may arise

2

• If the training dataset is not large enough, the data collected may not reflectthe complete property space. Consequently, many QSAR results cannot beused to confidently predict the most likely compounds of the best activity.

3

• There are many successful applications of this method but one cannot expect the QSAR approach to work well all the time. QSAR Models are easy to build but also very easy to get wrong.

QualityNano Training School 

Page 31: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Part IINano‐toxicity and

Nanomaterial Characterization31QualityNano Training School 

Page 32: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

No standardized or validated methods for nanotoxicity 

testing

Available toxicitydata for NPs areinconsistent and

confusing

Blocks the development of 

ENPs risk reduction strategies[2]

New Technology

New Concerns[1]

New Benefits

• Safety to human health and environment• Suitability of risk management strategies

Macro concerns with nano‐world !

32

[1] Chaudhry, Q. et al. (2010). Knowns, Unknowns, and Unknown Unknowns . RSC Nanoscience & Nanotechnology (p. 201-218).Royal Society of Chemistry.[2]Tran, L., & Chaudhry, Q. (2010). Engineered Nanoparticles and Food: An Assessment of Exposure and Hazard. RSC Nanoscience & Nanotechnology (p 120-134). Royal Society of Chemistry.

Macro concerns with nano‐world !

QualityNano Training School 

Uncertainities at the early stages of a new technology bring new concerns!

More than 800 nano‐products

Page 33: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

33

Risk assesment for NMS [1]

1.HazardIdentification

2.Doseresponse

relationship

3.Exposureassessment

4.Risk assessment

•What dangers to human health may basically arise from the harmful agent ?

•What is the nature and magnitude of the 

risk and how accurately can it be 

estimated?

•To what extent is the relevant 

population group exposed to the harmful agent?

•What quantitative connections existbetween the dose

and the extent of theexpected effect?

OECD (2003): Environment Directorate. Description of Selected Key Generic Terms Used in Chemical / Hazard Assessment. OECD Series on Testing and Assessment Number 44. ENV/JM/MONO(2003)15. Paris, 30-Oct-2003.

Risk assesment for NMS [1]

Potential safety issues of NMs should be addressed at the same time as thetech. is developing

QualityNano Training School 

ENMs are already used in a number of commercial applications.

Page 34: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

34

Hazard Identification

Dose‐response evaluationEvaluation of toxicity

involves determining the conditions or levels at which the presence of a contaminant may trigger a biological response from the body

Different amount ofCNT dust

At level of 0.1 mg/m3

At higher level of0.5 mg/m3

No effects wereobserved

Side effects weredetected in the

lungNaidu, B. David (2009). Biotechnology and Nanotechnology - Regulation under Environmental, Health, and Safety Laws.. Oxford University Press.

Dose‐response relationship

QualityNano Training School 

The evaluation of toxicity includes two steps:

Page 35: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

35

•The only reasonable way to obtain toxicity information for the large 

number of nanoparticles without testing every single one is to relate 

the physicochemical characteristics of NPs with their toxicity in a SAR 

(Structure Activity Relationship) or QSAR model.

•Nanomaterial characterization is one of the first steps in toxicity test.

However, it is a very challenging issue!!!

QualityNano Training School 

Particle Characterization

Nanoparticle Structure Properties

Particle SizeSurface AreaMorphology

State of Dispersion,…

Toxicity

Page 36: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Size and shape Size distribution Agglomeration state Porosity Structure‐dependent elect.  Electronic properties Characterisation: 

In what medium?

POSSIBLE factors[1]inducing toxicity of 

chemicals

Surface area Surface chemistry Surface charge Crystal structure Composition Configuration Coating Aggregation state Metal content

When a material is in a nano‐sized form, its properties become different from thesame material in a bulk form

Fullerenes Carbon‐based Different and             Carbon NTs Nanomaterials  Exceptional Graphene Properties

36[1] Puzyn, T., Leszczynska, D. & Leszczynski, J. Toward the development of ‘Nano-QSARs’: advances and challenges. Small 5, 2494–2509 (2009).

Possible toxicity‐related Physico‐chemical Properties

QualityNano Training School 

No certain information!!!Recent studies showed that toxicity of NPs can be related to:

more research is required to reach a clear understanding

more research is required to reach a clear understanding of the

Page 37: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Shape and aggregation state

Composition, purity and surface chemistry

Surface charge

Crystal structure

Particle number and size distribution

Surface Area

High Resolution Microscopy

Spectroscopy and Chromatography Methods

Zeta Potential Analysis

X‐Ray Diffraction

Several Techniques (e.g. SEM,TEM,DLS)

Brunauer‐Emmett‐Teller Adsorption

Several different methods and instruments are used to determine the properties on NMs

37

Nanomaterial Characterization

QualityNano Training School There is no single instrument that is right tool for every test. There are in fact more than 400 

different techniques for particle counting and characterization

Page 38: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

38

Some NPs 

Have a tendency to form clusters (e.g. aluminium oxide)

in an agglomeration state,some NPs behave like larger particles in the media

Toxicity may also depends on the size of agglomerate

NOT on original particle size itself !!!

Hussain, S. e. (2008). Can silver nanoparticles be useful as potential biological labels? Nanotechnology 19 .

If the NPs aggregateit is important to identify how these changes affect toxicity

How to characterize aggregation ?Texture analysis ?

What haven't we considered yet ?

Page 39: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

39

Challenges for QSAR

NP type

Size(nm)

Doses(mg/ml) Cell Method Result

Ag 7‐20 0.8‐50Human  fibrosorcoma and skin carcinoma 

XTT assay Non‐toxic Araro et al. (2008)[1]

Ag <15 0.7 Human hepatoma MTT Alamar blue and LDH assay

Oxidative stres mediated toxicity

Kim et al. (2009)[2]

Ag15 and100

10‐250Rat liver deliverd cellline

MTT, LDH, GSHlevel, ROS and MMP

Toxic Hussain et al. (2005)[3]

Differences in NMs used(size)

Differences inexperimentalconditions

[1] Arora S, Jain J, Rajwade JM, Paknikar KM (2008). Cellular responses induced by silver nanoparticles: in vitro studies. Toxicol Lett 179:93–100[2] Kim S, Choi JE, Choi J, Chung KH, Park K, Yi J, Ryu DY (2009). Oxidative stress-dependent toxicity of silver nanoparticles in humanhepatoma cells. Toxicol In Vitro 23:1076–1084[3] Hussain SM, Javorina AK, Schrand AM, Duhart HM, Ali SF, Schlager JJ (2006). The interaction of manganese nanoparticles with PC-12 cells induces dopamine depletion. Toxicol Sci 92:456–463

Challenges for nano‐QSAR ?

QualityNano Training School 

Table shows the results of 3 different toxicity studies for silver NMs:

strongly affectsthe results

Page 40: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

40

Challenges for nano‐QSAR [1]

[1] T. Puzyn et al. (eds.), Recent Advances in QSAR Studies, 103–125. DOI 10.1007/978‐1‐4020‐9783‐6_4, C  Springer Science+Business Media B.V. 2010

1• Lack of high‐quality experimental data

2• Lack of conceptual frameworks for grouping NPs according to mode of physicochemical properties and toxic action

3• Lack of sufficient molecular descriptors able to express specificity of “nano” structure

4• Lack of rational modeling procedures to screen large numbers ofstructurally diversified nanoparticles.

5• Limited knowledge on the interactions between NPs and biological systems

QualityNano Training School 

NMs are problematic for QSAR modellers because of the several reasons:

Page 41: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Part IIICase Study

despite all limitations…41QualityNano Training School 

Page 42: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Nanotxicology 2012, 4‐7 September, Beijing

The aim of this study was to test the hypothesis that nanoparticle

toxicity is a function of some measurable physical characteristics and

that Structure Activity Relationship (SAR) might ultimately be a useful

tool for nanotoxicology. A panel of 18 nanoparticles was chosen to

include carbon‐based materials and metal oxides. Comparative toxicity

of the nanoparticles in the panel was assessed, using a number of

different measures of apoptosis and necrosis; haemolytic effects and

the impact, on cell morphology were also assessed. Structural and

composition properties were measured including size distribution,

surface area, morphological parameters, metal concentration,

reactivity, free radical generation and zeta potential. Toxicity end

points were processed using principal component analysis and

clustering algorithms.

QSAR Toxicity (in vitro) Analysis for a Panel of Eighteen NPs

42QualityNano Training School 

Page 43: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Nanooxicology 2012, 4‐7 September, Beijing

Particles Qualifier Number Supplier Form

Carbon black Printex 90 N1 Degussa DryDiesel exhaust particles EPA N2 EPA Dry

Nanotubes Japanese N3 Vicki Stone DryFullerene N4 Sigma Dry

Polystyrene latexUnmodified N5 Polysciences Sus.

Amine N6 Sigma Sus.Carboxylated N7 Polysciences Sus.

Aluminium oxide7nm N8 Krahn Chemie Dry50nm N9 AHT* Dry300nm N10 AHT* Dry

Cerium oxide N11 NAM** DryNickel oxide N12 NAM** DrySilicon oxide N13 NAM** DryZinc oxide N14 NAM** Dry

Titanium dioxideRutile N15 NAM** DryAnatase N16 NAM** Dry

SilverN17N18

NAM**NAM**

DrySus.* AHT: Allied High Tech, ** NAM: Nanostructure and Amorphous Material Inc

The Set of Eighteen NPs

43QualityNano Training School 

Page 44: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Nanotoxiclogy 2012, 4‐7 September, Beijing

•Particle shape was analysed using LEO 1530 Scanning Electron Microscope (SEM) or PhilipsCM20 Transmission Electron Microscope (TEM) aspect ratio and mean size•Surface area and porosity were measured using TriStar 3000 BET. Thirteen measurements,including five surface areas based on different definitions, three pore volumes, three poresizes as well as the mean size and particle density.•Particle size and size distribution were analysed using a Malvern MasterSizer 2000. Sizedistribution, and seven other size properties (mass diameter, uniformity, specific surfacearea, surface area mean diameter and three mass diameters)•The free radical activities were measured by EPR (Electron paramagnetic resonance)using the spin traps, DMPO and Tempone H separately. Tempone H has partial selectivity fortrapping superoxide anions, whilst DMPO traps mainly hydroxyl radicals•Particle reactivity in solution, the dithiothreitol (DTT) consumption test as a reducing specieswas carried out. Since the DTT consumption test can only be conducted on dry powders, onlyfourteen of the panel of nanoparticles were assessed using this assay.

•Metal Concentration water soluble concentration of ten heavy metals (Ti, V, Cr, Mn, Fe, Co,Ni, Cu, Zn and Cd) was measured•Charge: z potential was measured using Malvern Instrument’s Zetasizer Nano instrument

Characterisation of  Physico‐chemical  Properties

44QualityNano Training School 

Page 45: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

45

1313 • BET measurements (for 14 dry samples)

22 • SEM/TEM measurements (for all samples)

77 • laser diffraction size statistical measurements (Master Sizer 2000) for all samples

8484 • laser diffraction size distribution measurements

22 • EPR measurements (for all samples)

11 • reactivity measurement (for dry samples)

1010 • metal content measurements

Total Number of Measurements

Total:119 Measurements

Particlename

BET Measurements M2000 Measurements

SEM Measurements Zeta measurements Other Measurements

N1 A(1,1),A(1,2),…,A(1,12) B(1,1),B(1,2)… C(1,1),C(1,2)… D(1,1),D(1,2)… E(1,1),E(1,2)…

N2 A(2,1),A(2,2),…,A(2,12 B(2,1),B(2,2)… C(2,1),C(2,2)… D(2,1),D(2,2)… E(2,1),E(2,2)…

… … … … … …N18 A(18,1),A(17,2),… B(18,1),B(18,2)… C(18,1),C(18,2)… D(18,1),D(18,2)… E(18,1),E(18,2)…

QualityNano Training School 

Page 46: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Nanotoxicolgy 2012, 4‐7 September, BeijingSEM and TEM images of the 18 nanoparticle samples

46QualityNano Training School 

Page 47: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Table 2. The water soluble concentrations of various metals for the 18 nanoparticle samplesNanotoxicology 2012, 4‐7 September, Beijing

Particle names Metal Concentration (µg/g)Ti V Cr Mn Fe Co Ni Cu Zn Cd

Carbon Black N1 0.000 0.057 0.000 0.305 0.000 0.002 0.263 1.161 9.687 0.008

Diesel Exhaust N2 2.320 0.312 16.47 20.81 208.4 0.508 4.940 2.235 3181 0.387

Japanese Nanotubes N3 0.000 0.100 0.000 0.161 13.10 0.000 0.299 0.123 0.460 0.001

Fullerene N4 0.000 0.084 0.000 0.000 0.000 0.000 0.534 0.142 4.837 0.000

Polystyrene Latex Beads N5 0.000 1.172 0.000 0.000 0.000 0.000 0.000 0.625 19.56 0.000

Polystyrene Latex Beads N6 4.368 0.779 0.343 0.199 0.000 0.006 0.440 18.23 58.71 0.010

Polystyrene Latex Beads N7 0.000 1.018 0.000 0.000 0.000 0.000 0.000 2.588 22.323 0.000

Aluminuim Oxide N8 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.714 0.000 0.000

Aluminuim Oxide N9 0.000 0.031 0.000 0.000 0.000 0.000 0.056 0.099 4.987 0.000

Aluminuim Oxide N10 0.000 0.062 0.000 0.000 0.000 0.000 0.023 0.088 0.000 0.000

Cerium Oxide N11 0.000 0.004 1.058 0.128 0.000 0.000 0.616 0.191 112.9 0.007

Nickel Oxide N12 0.000 0.056 0.000 0.000 0.000 0.042 16963 0.286 0.000 0.038

Silicon Oxide N13 0.000 0.088 0.000 0.000 0.000 0.000 14.163 0.106 0.000 0.000

Zinc Oxide N14 0.000 0.096 0.000 0.000 0.000 0.000 1.336 0.047 8185 9.711

Titanium Dioxide Rutile N15 0.000 0.083 0.000 0.000 0.000 0.000 0.286 0.276 39.623 0.000

Titanium Dioxide Anatase N16 0.000 0.104 0.000 0.160 0.000 0.006 0.692 0.303 402.3 0.000

Silver N17 0.000 0.111 0.000 0.503 0.000 0.000 0.475 0.597 17.867 0.019

Silver N18 0.000 0.096 0.000 0.256 0.000 0.059 0.561 162.8 194.8 0.037

The heavy metal concentration of each NP

47QualityNano Training School 

Page 48: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

LEO 1530 Gemini FEGSEMScanning Electron Microscopy

The information about Size, Shape can be obtained from SEM images

Carbon Black N1

Nanotoxicology 2012, 4‐7 September, Beijing

Mean size obtained by measuring the particles on the SEM/TEM image: Mean_Size_SEMAspect ratio obtained from SEM/TEM images: Aspect_Ratio

Characterisation of Particle Size and Shape

48QualityNano Training School 

Page 49: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Nanotoxicology 2012, 4‐7 September, Beijing

Micromeritics TriStar 3000Surface Area Technique: Gas AdsorptionMinimum Surface Area:     0.01m2/g

Support Gases:     Nitrogen and HeliumMeasurement Parameters: BET Measurement

Adsorption IsothermsDesorption Isotherms

t‐plotsPore Volume AnalysisPore Size Analysis

Surface Area (m2/g) Pore Volume (cm3/g) Pore Sizes (Å)

Single point surface area at P/Po =0.197 (m²/g) – SPSA

BET Surface Area: BET_SA Langmuir Surface Area: LM_SA BJH Adsorption cumulative surface

area of pores between 17.000 Å and3000.000 Å diameter :BJH_AC_SAOP

BJH Desorption cumulative surfacearea of pores between 17.000 Å and3000.000 Å diameter:BJH_DC_SAOP

Single point adsorption total pore volume of pores at P/Po = 0.98: SP_TP_VOP

BJH Adsorption cumulative volume of pores between 17.000 Å and 3000.000 Å diameter: BJH_AC_VOP

BJH Desorption cumulative volume of pores between 17.000 Å and 3000.000 Å diameter : BJH_DC_VOP

Adsorption average pore width: BET_PW

BJH Adsorption average pore diameter : BJH_A_PD

BJH Desorption average pore diameter: BJH_D_PD

Mean size calculated from BET surface area: Mean_Size_BET

Particle density: Density

** 14 dry powder samples, no available for three suspensions (N5,N6, and N7) and the new Silver sample.

BET Measurements for 14 dry powder samples

49QualityNano Training School 

Page 50: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Malvern Mastersizer 2000Type of Analyser: Laser Diffraction

Sample Volume: 1000ml, 150ml (Hydro G and S)Measurement Range: 20nm – 2000μm

0123456789

10

0.01 0.1 1 10 100 1000 10000

Size (micron)

Prob

abili

ty (%

)

The information about mean Size and Size distribution, can be obtained from this sizer. 

Nanotoxicology 2012, 4‐7 September, BeijingCharacterisation of NPs‐Mastersizer

50QualityNano Training School 

Page 51: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

D[4,3] – the Volume Weighted Mean or Mass Moment Mean Diameter, represent by D[4,3]

Uniformity: a measure of the absolute deviation from the median, represented by Uniformity

Specific surface area: the total area of the particles divided by the total weight, SSA_2000

D[3,2] – the Surface Weighted Mean or Surface Area Moment Mean Diameter, represented by D[3,2]

D(v,0.5) – the size in microns at which 50% of the sample is smaller and 50% is larger. This value is also known as the Mass Median Diameter or the median of the volume distribution, represented by D(0.5)

D(v,0.1) – the size of particle below which 10% of the sample lies, represented by D(0.1)

D(v,0.9) – the size of particle below which 90% of the sample lies, represented by D(0.9)

Size distribution: there are 100 values. Among them there are 16 values equal to zero for all particles. So 84 size distribution attributes are used for further analysis

Mastersizer 2000 Measurements: 91**

** Note: available for all samples

Nanotoxicology 2012, 4‐7 September, BeijingCharacterisation of NPs‐Mastersizer

51QualityNano Training School 

Page 52: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Malvern ZetaSizerNanoinstrument Size ranges: 0.6nm ‐ 6µm

Zeta Potential Suitable: suspensions

Nanotoxicology 2012, 4‐7 September, BeijingCharacterisation of NPs‐Zeta Potential

52QualityNano Training School 

Page 53: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Cytotoxicity by LDH Release

N6

N12

0.00

20.00

40.00

60.00

80.00

100.00

31.25 62.50 125 250

LDH Release

Dose (g/ml)

N6

N14

0.00

5.00

10.00

15.00

20.00

25.00

31.25 62.50 125 250

Apop

tosis

Dose (g/ml)

Apoptosis

N6

N14

0.00

20.00

40.00

60.00

80.00

100.00

120.00

31.25 62.50 125 250

Viab

le

Dose (g/ml)

Viability

N3

N6

N14

0

10

20

30

40

50

60

70

80

31.25 62.50 125 250

Necrosis

Dose (g/ml)

Necrosis

0

10

20

30

40

50

60

70

80

90

100

Haemolysos at d

ose 50

0  g/ml

Pro‐inflammation effects Haemolysis

53QualityNano Training School

Toxicity Results

Page 54: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

54

LDH Apoptosis Viability

NecrosisHaemolysis

Total‐ToxicityMTT

QualityNano Training School 

Principal Component Analysis of Cytotoxicity Data

Page 55: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

55

Figure : Plot of the first and second  PCs based on PCA analysis of toxicity data 

Japanese nanotubes

Zinc oxide

Nickel oxide

Polystrene larex (Amine)

‐ The major cluster contains low toxicity particle 

samples  

‐Aminated beads (N6), zinc oxide (N14), Japanese 

Nanotubes (N3) and nickel oxide (N12) are separate 

from the major cluster. ‐ Aminated beads (N6) has the highest toxicity values in nearly all assay results (LDH, apoptosis, necrosis, haemolytic, MTT and cell morphology assays).

‐ Zinc oxide (N14) has high toxicity value in LDH, 

apoptosis, necrosis and inflammation assays. 

‐ Japanese nanotubes (N3) have high toxicity values 

in viability and MTT assays. 

‐Nickel oxide (N12) has shown highly toxic in LDH 

and haemolytic assays results..

QualityNano Training School 

Principal Component Analysis of Cytotoxicity Data

The distinction of the four particles, N6, N14, N3 and N12, from other particles is clear.

Page 56: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

56

NP6‐Average Toxicity Comparison NP14‐Average Toxicity Comparison

N14 has high toxicity value in LDH, apoptosis,necrosis and inflammation assays.

N12 has shown highly toxic in LDH and  haemolyticassays results 

NP12‐Average Toxicity Comparison NP3‐Average Toxicity Comparison

N3 have high toxicity values in viability and MTT assays. 

N6 has the highest toxicity values in nearly all assayresults (LDH, apoptosis, necrosis,  haemolytic, MTT and cell morphology assays). 

QualityNano Training School The toxicity results of N6, N3, N12 and N14 which are likely to be toxic

Page 57: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

57

The focus of the (Q)SAR Analysis is to identify the possible structural andcompositional properties which may contribute to the toxicity of zincoxide (N14), polystrene latex amine (N6), Japanese nanotubes (N3) andnickel oxide (N12).

Next Step: Principal Component Analysis of Descriptors

PCA and clustering were then applied to the physicochemicaldescriptors with the hypothesis that the toxic nanoparticles might alsobe discriminated from the non‐toxic particles based on the analysis ofthe measured physicochemical descriptors.

The theory is that there should be structural or/and compositionaldistinctions for these nanoparticles from the rest of the non‐toxicparticles.

If this can be proved, further analysis can be conducted to identify thekey physicochemical descriptors that lead to the observed toxicity.

QualityNano Training School 

Page 58: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

58

N2

N3

N12

N14

N6

N5

N7

Figure : PCA analysis of the physicochemicaldescriptors for the panel of 18 nanoparticles(excluding BET and DTT measured descriptorssince they are not available for the four wetsamples).

QualityNano Training School 

Principal component analysis of the structural and compositional descriptors for the  panel of 18 nanoparticles (excluding BET and DTT measured descriptors since they are not available for the four wet samples). 

Principal Component Analysis of Descriptors

‐The scaling operation was performed, 3PCs 

were then applied.

‐PC1‐PC2‐PC3 three dimensional plot shows 

that the 18 nanoparticles were clearly grouped 

into clusters.

‐The largest cluster contains particles that 

showed low toxicity values.

‐Japanese Nanotubes (N3), nickel oxide (N12) 

and zinc oxide (N14) and diesel exhaust 

particles (N2) are outside this low toxicity 

sample cluster.

‐There must be another reason that leads to 

high toxicity value for N6

Page 59: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

59

•To examine the clustering results more clearly, PC1‐PC2,  PC1‐PC3 and PC2‐PC3 two dimensional plots are also examined .•It is PC2 that separates N3(Japanese nanotubes), N12 (nickel oxide) and N14 (zinc oxide)  from the rest samples, as can be seen from PC1‐PC2 and PC2‐PC3 plots.

QualityNano Training School 

Results

Page 60: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

60

Figure: PC1, PC2 and PC3 Contribution plots based on PCA analysis of the structural and compositional descriptors for the panel of 18 nanoparticles

QualityNano Training School 

Results

•The contribution plot for, PC1, PC2 and

PC3 is shown in the Figure, we are mainly

interested in the contribution to PC2 (solid

red colour), since it is PC2 that

discriminated Japanese nanotubes (N3)

nickel oxide (N12) and zinc oxide (N14)

from the rest particle samples.

•The largest contributions to PC2 are: aspect

ratio measured by SEM imaging, and volume

weighted mean, uniformity, and D(0.9), all

measured by Mastersizer, as well as Ni and

Zn metal contents

•Further analysis is required.

Page 61: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

61

N2

N 3

N 12

N14

•In order to determine exactly whichphysicochemical descriptors lead to high toxicity values of N3, N12 ,N14and N6, several PC analysis wereperformed.

•For the 14 dry particles, PC analysis with BET and DTT measurements gavesimilar result with the first PC analysisin which BET and DTT data wasexcluded.

QualityNano Training School 

Figure : Principal component analysis of the structural and compositional descriptors (all such descriptors including BET and DTT measurements) for the 14 dry nanoparticles.

Results

Page 62: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

62

Figure: PC1 and PC2 contribution plots for PCA analysis of  the compositional descriptors

• Ni content made the most important contribution to PC1, suggesting that nickel content may be the main reason for the high toxicity of N12 (nickel oxide). • While Zn content made the most obvious contribution to PC2, suggesting  that zinc content may be the reason for the high toxicity of N2 (diesel exhaust particles) and N14 (zinc oxide).

QualityNano Training School 

Results

•To identify the relativecontribution of each metalconcentration to PC1 andPC2, the contribution plot toPC1 and PC2 is shown in theFigure.

Page 63: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

Table 2. The water soluble concentrations of various metals for the 18 nanoparticle samplesNanotoxicology 2012, 4‐7 September, Beijing

Particle names Metal Concentration (µg/g)Ti V Cr Mn Fe Co Ni Cu Zn Cd

Carbon Black N1 0.000 0.057 0.000 0.305 0.000 0.002 0.263 1.161 9.687 0.008

Diesel Exhaust N2 2.320 0.312 16.47 20.81 208.4 0.508 4.940 2.235 3181 0.387

Japanese Nanotubes N3 0.000 0.100 0.000 0.161 13.10 0.000 0.299 0.123 0.460 0.001

Fullerene N4 0.000 0.084 0.000 0.000 0.000 0.000 0.534 0.142 4.837 0.000

Polystyrene Latex Beads N5 0.000 1.172 0.000 0.000 0.000 0.000 0.000 0.625 19.56 0.000

Polystyrene Latex Beads N6 4.368 0.779 0.343 0.199 0.000 0.006 0.440 18.23 58.71 0.010

Polystyrene Latex Beads N7 0.000 1.018 0.000 0.000 0.000 0.000 0.000 2.588 22.323 0.000

Aluminuim Oxide N8 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.714 0.000 0.000

Aluminuim Oxide N9 0.000 0.031 0.000 0.000 0.000 0.000 0.056 0.099 4.987 0.000

Aluminuim Oxide N10 0.000 0.062 0.000 0.000 0.000 0.000 0.023 0.088 0.000 0.000

Cerium Oxide N11 0.000 0.004 1.058 0.128 0.000 0.000 0.616 0.191 112.9 0.007

Nickel Oxide N12 0.000 0.056 0.000 0.000 0.000 0.042 16963 0.286 0.000 0.038

Silicon Oxide N13 0.000 0.088 0.000 0.000 0.000 0.000 14.163 0.106 0.000 0.000

Zinc Oxide N14 0.000 0.096 0.000 0.000 0.000 0.000 1.336 0.047 8185 9.711

Titanium Dioxide Rutile N15 0.000 0.083 0.000 0.000 0.000 0.000 0.286 0.276 39.623 0.000

Titanium Dioxide Anatase N16 0.000 0.104 0.000 0.160 0.000 0.006 0.692 0.303 402.3 0.000

Silver N17 0.000 0.111 0.000 0.503 0.000 0.000 0.475 0.597 17.867 0.019

Silver N18 0.000 0.096 0.000 0.256 0.000 0.059 0.561 162.8 194.8 0.037

The heavy metal concentration of each NP

63QualityNano Training School 

Page 64: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

64

Identifying the factor that discriminates N6 (polystrene latex amine) from N5 (unmodified) and N7 (carboxylated) 

PCA analysis:  PCA analysis of metal contents only, or  PCA analysis of structural descriptors only, or PCA analysis of both structural and compositional descriptors together, could not discriminate N5, N6 and N7 from the largest clusters of non‐toxic particle samples. 

PCA Conclusion: neither structural descriptors nor metal contents made N6 different fro N5 and N7 in  toxicity

It was at this stage of analysis that we suspected that there are other possible reasons for toxicity specific to N6: either it was due to the charge difference, or due to N6 being amine, or both. 

zeta potential for all three polystyrene latex samples.N6 (amine)  37.8, 37.5, and 40.3 (mV),    positive N5 (unmodified)  ‐36.2, ‐38.8 and ‐36.8,  negativeN7 (carboxylated) ‐54.9, ‐55.3, and ‐58.6., negative

According to Verma et al. any particle with positive charge is likely to interact electrostatically since most cell surfaces and other biological membranes are negatively charged under physiological conditions(Verma et al. 2008).  Therefore, it is most likely that the large positive charge of N6 contributed its observed high toxicity, despite it structurally similar to N5 and N7.

Conclusions

QualityNano Training School 

Page 65: “Quantitative Structure Activity Relationship (QSAR) …qualitynano.eu/.../7.QSARs-Lecture_CeydaOksel.pdfQSAR (Quantitative Structure‐Activity Relationship) nano‐QSAR QSAR models

65

• The SAR analysis generates the following conclusions:1. The most likely reason of high toxicity for Japanese nanotube (N3) is its shape (high 

aspect ratio). 2. The most likely cause of high toxicity for zinc oxide (N14) and nickel oxide (N12) is 

their high contents of zinc and nickel. 3. Although, Aminated beads (N6) has the highest toxicity values in nearly all assay 

results , it was always grouped closely with the other two polystyrene latex beads, unmodified beads (N5) and carboxylated beads (N7), which showed no toxicity. Therefore, the zeta potential of the three polystrene latex was measured. The result suggests that the large positive charge of N6 contributed to its observed high toxicity.

• The work has shown that SAR based on molecular descriptors is a usefultool for describing factors influencing the toxicity of nanoparticles. 

• A panel of eighteen nanoparticles is still considered to be too small, therefore, a QSAR model that can be applied to predict the toxicity of new nanoparticles cannot be developed with this small set of data.

QualityNano Training School 

Conclusions