categorization of uvcbs using chemical-biological read across · 2019-08-26 · categorization of...

1
CATEGORIZATION OF UVCBS USING CHEMICAL-BIOLOGICAL READ ACROSS Grimm FA 1 , Iwata Y 1 , Sirenko O 2 , Russell WK 1 , Luo Y 1 , Crittenden C 2 , Wright FA 3 , Reif DM 3 , Yeakley J 4 , Seligmann B 4 , Shepard PJ 4 , Roy T 5 , Boogaard PJ 6 , Ketelslegers H 7 , Rohde AM 7 , and Rusyn I 1 1 Texas A&M University, College Station, TX, USA; 2 Molecular Devices LLC, Sunnyvale, CA, USA; 3 North Carolina State University, Raleigh, NC, USA; 4 Biospyder Technologies Inc., Carlsbad, CA; 5 University of South Carolina, Beaufort, SC, USA; 6 SHELL International BV, The Hague, NL; and 7 Concawe, Brussels, BE RATIONALE OBJECTIVES CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES RESULTS Curve-fitting of quantitative outputs from HCS applications (A) (shown are representative dose-response curves for cardiomyocyte beat frequency and hepatocyte viability) in R yielded point-of-departure (POD) benchmark values, relative measures of toxicity that were integrated for quantitative bioactivity profiling in ToxPi (B). 7-9 Each ToxPi slice is representative of an individual assay and the area of the slice is proportional to the relative bioactivity of the chemical within the dataset. Computational Analysis of IM-MS Data Provides Unique Feature Identification MA plots show the change in the average level of each transcript of the S1500+ gene set for a representative straight-run gas oil (CON-02) and a heavy fuel oil (A087/13) relative to the respective averages determined for DMSO controls (A). Although the specific differentially expressed genes varied among the different petroleum substances, the overall treatment effect was found to be correlated (r 2 =0.49) (B). Principal components analysis of global changes in gene expression revealed a clustering trend of petroleum substances (C). Download poster PDF from the following URL: Chemicals of Unknown or Variable composition, Complex reaction products, and Biological materials (UVCBs) present a major challenge for registrations under the REACH and US High Production Volume regulatory programs. In addition to their complex chemical composition, gaps in available toxicity data underline the need for confident categorization of these substances to allow read across applications. Here, we present a comprehensive experimental and computational approach to categorize UVCBs according to global similarities in (1) their chemical composition using Ion Mobility Mass Spectrometry (IM-MS) and (2) their bioactivities using a suite of in vitro models. For chemical read across, we analysed 18 petroleum substances from four distinct product groups by IMMS to determine substance-specific quantitative parameters including m/z distribution, drift time, carbon numbers, and double bond equivalents. For biological read across, we exposed induced pluripotent stem cell-derived cardiomyocytes and hepatocytes to a DMSO-soluble extract series of 21 petroleum substances comprising five product groups for up to 72 hours. WORKFLOW ION MOBILITY-MASS SPECTROMETRY PHENOTYPIC IN VITRO SCREENING Differential Gene Expression Analysis in petroleum substance-treated iPSC hepatocytes HIGH-THROUGHPUT TRANSCRIPTOMICS Determine “fingerprints” of petroleum substances using computational analysis of IM-MS data sets Collect bioactivity data on in vitro effects of petroleum substances using quantitative high-content screening of iPSC-derived cardiomyocytes and hepatocytes Determine the effects of petroleum substance extracts on differential gene expression in iPSC hepatocytes using the TempO-seq transcriptomics assay Address the challenge of safety assessment of UVCB materials through “chemical-biological” data- integrative groupings using a case study of petroleum substances Quantitative Dose-Response Data Global ranking of petroleum substances according to their cumulative ToxPi score, i.e. each data point represents the sum of individual assay scores shown in the respective bioactivity profiles. Bioactivity-based correlations were analyzed using the hclust function in R. Combinatorial integration of physico-chemical and bioactivity data for groupings of petroleum substances in ToxPi was achieved through parameter-specific clustering of assay parameters (A). Correlation between physico-chemical data-derived ToxPi scores (data not shown) and bioactivity data-derived ToxPi scores is shown in panel (B). Chemical Composiiton-Based Grouping of Petroleum Substances Chemical Fingerprinting and Categorization of Petroleum Substances Biological Data-Integrative Categorization of Petroleum Substances Data Integration and Bioactivity Profiling in ToxPi Increasing the Confidence in Petroleum Substance Categories Using Chemical-Biological Data IMMS of concawe samples 1,2 : Petroleum substances were diluted to 1mg/mL in a 1:1 Methanol:Toluene in 0.5% formic acid solution and infused for a duration of 2 min into a Waters G2 synapt Q-TOF MS. The ion mobility wave velocity was set to 800 m/s and the wave height set to 40V. Petroleum Extract Preparation: DMSO soluble extracts of petroleum substances from five distinct manufacturing categories (SRGO - Straight Run Gas Oils, OGO - Other Gas Oils, VHGO - Vacuum & Hydrotreated Gas Oils, RAE - Residual Aromatic Extracts, and HFO - Heavy Fuel Oils) were provided by Concawe (Brussels, Belgium). Cell Culture 3-6 : iCell Cardiomyocytes and iCell Hepatocytes 2.0 were purchased from Cellular Dynamics International (CDI) and cultured according to the manufacturer’s protocols. In Vitro Cardiotoxicity Assay 3-5 : Effects on cardiophysiology were assessed by monitoring the intracellular Ca 2+ -flux of synchronously contracting cardiomyocytes using the FLIPR Tetra system (Molecular Devices LLC). High-Content Imaging 3,6 : Cytotoxicity screening after 24 (cardiomyocytes) or 72 (hepatocytes) hours of treatment with petroleum substance extracts was performed using the ImageXPress Micro XL high-content imaging system (Molecular Devices) following staining with fluorescent probes Hoechst 33342, Calcein AM, MitoTracker Orange, and AF488-conjugated phalloidin. TempO-seq assay: Global changes in gene expression patterns in hepatocytes treated with petroleum extracts for 48 hours were analyzed using a targeted RNA sequencing technology, TempO-Seq™ (BioSpyder Technologies, Inc., Carlsbad, CA) measuring a set of genes selected by an effort organized by NIEHS to represent a surrogate whole transcriptome assay, the S1500. Differentially expressed genes were identified using the DESeq2 R package. Petroleum substances, prototypical high-production volume UVCBs, can be categorized using global similarities in their bioactivity profiles using multi-parametric HCS of iPSC-derived cell types In combination with transcriptomics analysis, interpretation of these multidimensional data sets is not limited to biological property-derived groupings for regulatory applications, but may eventually also be informative for mechanistic toxicity evaluations Physico-chemical descriptors and global compositional analysis using Ion Mobility-Mass Spectrometry further improve the confidence in biological data-based groupings Together, our findings strengthen the argument that a combinatorial approach using quantitative chemical analysis, high-content in vitro screenings, and subsequent computational data integration and visualization possesses the potential to improve chemical-biological read-across applications MATERIALS & METHODS IM-MS spectra for each two representative Other Gas Oils (OGO), Straight Run Gas Oils (SRGO), Vacuum & Hydrotreated Gas Oils (VHGO), and Heavy Fuel Oils (HFO) (A). Computational integration of IM-MS data sets in PetroOrg software 2 provides unique feature identification and heteroatom class distribution for each sample (B). The upper panel depicts the total number of features for each individual sample. Averages for the top 10 most abundant heteroatom classes are shown in the lower panel. 1 Michelmann et al. (2014) J Am Soc Mass Spectrom. 26: 14-24 2 www.petroorg.com 3 Grimm et al. (2015) Assay Drug Dev Technol. 13: 529-546 4 Sirenko et al. (2013) Toxicol Appl Pharmacol. 273: 500-507 5 Sirenko et al. (2013) J Biomol Screen. 18: 39-53 6 Sirenko et al. (2015) Assay Drug Dev Technol. 12: 43-52 7 https://federalregister.gov/a/2015-08529 8 Robin et al. (2011) BMC Bioinformatics. 12: 77 9 US EPA (2011) Benchmark Dose Technical Guidance 10 Reif et al. (2013) Bioinformatics. 29: 402-403 QR-CODE The authors appreciate useful discussions and technical support from Joel McComb (BioSpyder Technologies Inc., Carlsbad, CA), Yi-Hui Zhou (North Carolina State University), Grace Chappell (Texas A&M University), John Braisted, and David Gerhold (National Institutes of Health/ National Center for Advancing Translational Sciences, Bethesda, MD). Petroleum substances were provided by Concawe, Brussel, BE). This work was supported by EPA STAR grant #RD83516601 and institutional support from Texas A&M University. Fabian Grimm is a recipient of SOT Colgate-Palmolive postdoctoral fellowship. FUTURE WORK A B Plots visualize unique feature distributions of the (most abundant) N1 heteroatom class, i.e. carbon chain length vs double bond equivalents (DBE) for each two representative OGOs, SRGOs, VHGOs, and HFOs. A B C Heatmap representation (A), cluster analysis (B), and principal components analysis (C) of relative abundances of heteroatom classes using the hclust function in R revealed substance and group-specific fingerprints of petroleum substances. A B Extensions of both chemical analysis and in vitro screenings to additional petroleum manufacturing classes Inclusion of a more comprehensive set of tissue types, i.e. iPSC-derived neurons, macrophages, endothelial cells, and skeletal myoblasts Estimating the interindividual and population variablity in phenotypic responses A B

Upload: others

Post on 20-Jun-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: CATEGORIZATION OF UVCBS USING CHEMICAL-BIOLOGICAL READ ACROSS · 2019-08-26 · CATEGORIZATION OF UVCBS USING CHEMICAL-BIOLOGICAL READ ACROSS Grimm FA1, Iwata Y1, ... Download poster

CATEGORIZATION OF UVCBS USING CHEMICAL-BIOLOGICAL READ ACROSS Grimm FA1, Iwata Y1, Sirenko O2, Russell WK1, Luo Y1, Crittenden C2, Wright FA3, Reif DM3, Yeakley J4, Seligmann B4, Shepard PJ4, Roy T5, Boogaard PJ6, Ketelslegers H7, Rohde AM7, and Rusyn I1

1Texas A&M University, College Station, TX, USA; 2Molecular Devices LLC, Sunnyvale, CA, USA; 3North Carolina State University, Raleigh, NC, USA; 4Biospyder Technologies Inc., Carlsbad, CA; 5University of South Carolina, Beaufort, SC, USA; 6SHELL International BV, The Hague, NL; and 7Concawe, Brussels, BE

RATIONALE

OBJECTIVES

CONCLUSIONS

ACKNOWLEDGEMENTS

REFERENCES

RESULTS

Curve-fitting of quantitative outputs from HCS applications (A) (shown are representative dose-response curves for cardiomyocyte beat frequency and hepatocyte viability) in R yielded point-of-departure (POD) benchmark values, relative measures of toxicity that were integrated for quantitative bioactivity profiling in ToxPi (B).7-9 Each ToxPi slice is representative of an individual assay and the area of the slice is proportional to the relative bioactivity of the chemical within the dataset.

Computational Analysis of IM-MS Data Provides Unique Feature Identification

MA plots show the change in the average level of each transcript of the S1500+ gene set for a representative straight-run gas oil (CON-02) and a heavy fuel oil (A087/13) relative to the respective averages determined for DMSO controls (A). Although the specific differentially expressed genes varied among the different petroleum substances, the overall treatment effect was found to be correlated (r2=0.49) (B). Principal components analysis of global changes in gene expression revealed a clustering trend of petroleum substances (C).

Download poster PDF from the following URL:

Chemicals of Unknown or Variable composition, Complex reaction products, and Biological materials (UVCBs) present a major challenge for registrations under the REACH and US High Production Volume regulatory programs. In addition to their complex chemical composition, gaps in available toxicity data underline the need for confident categorization of these substances to allow read across applications. Here, we present a comprehensive experimental and computational approach to categorize UVCBs according to global similarities in (1) their chemical composition using Ion Mobility Mass Spectrometry (IM-MS) and (2) their bioactivities using a suite of in vitro models. For chemical read across, we analysed 18 petroleum substances from four distinct product groups by IMMS to determine substance-specific quantitative parameters including m/z distribution, drift time, carbon numbers, and double bond equivalents. For biological read across, we exposed induced pluripotent stem cell-derived cardiomyocytes and hepatocytes to a DMSO-soluble extract series of 21 petroleum substances comprising five product groups for up to 72 hours.

WORKFLOW

ION MOBILITY-MASS SPECTROMETRY PHENOTYPIC IN VITRO SCREENING

Differential Gene Expression Analysis in petroleum substance-treated iPSC hepatocytes

HIGH-THROUGHPUT TRANSCRIPTOMICS

Determine “fingerprints” of petroleum substances using computational analysis of IM-MS data sets Collect bioactivity data on in vitro effects of petroleum substances using quantitative high-content

screening of iPSC-derived cardiomyocytes and hepatocytes Determine the effects of petroleum substance extracts on differential gene expression in iPSC

hepatocytes using the TempO-seq transcriptomics assay

Address the challenge of safety assessment of UVCB materials through “chemical-biological” data-integrative groupings using a case study of petroleum substances

Qu

anti

tati

ve D

ose

-Re

spo

nse

Dat

a

Global ranking of petroleum substances according to their cumulative ToxPi score, i.e. each data point represents the sum of individual assay scores shown in the respective bioactivity profiles. Bioactivity-based correlations were analyzed using the hclust function in R.

Combinatorial integration of physico-chemical and bioactivity data for groupings of petroleum substances in ToxPi was achieved through parameter-specific clustering of assay parameters (A). Correlation between physico-chemical data-derived ToxPi scores (data not shown) and bioactivity data-derived ToxPi scores is shown in panel (B).

Chemical Composiiton-Based Grouping of Petroleum Substances

Chemical Fingerprinting and Categorization of Petroleum Substances

Biological Data-Integrative Categorization of Petroleum Substances

Data Integration and Bioactivity Profiling in ToxPi

Increasing the Confidence in Petroleum Substance Categories Using Chemical-Biological Data

IMMS of concawe samples1,2: Petroleum substances were diluted to 1mg/mL in a 1:1 Methanol:Toluene in 0.5% formic acid solution and infused for a duration of 2 min into a Waters G2 synapt Q-TOF MS. The ion mobility wave velocity was set to 800 m/s and the wave height set to 40V.

Petroleum Extract Preparation: DMSO soluble extracts of petroleum substances from five distinct manufacturing categories (SRGO - Straight Run Gas Oils, OGO - Other Gas Oils, VHGO - Vacuum & Hydrotreated Gas Oils, RAE - Residual Aromatic Extracts, and HFO - Heavy Fuel Oils) were provided by Concawe (Brussels, Belgium).

Cell Culture3-6: iCell Cardiomyocytes and iCell Hepatocytes 2.0 were purchased from Cellular Dynamics International (CDI) and cultured according to the manufacturer’s protocols.

In Vitro Cardiotoxicity Assay3-5: Effects on cardiophysiology were assessed by monitoring the intracellular Ca2+-flux of synchronously contracting cardiomyocytes using the FLIPR Tetra system (Molecular Devices LLC).

High-Content Imaging3,6: Cytotoxicity screening after 24 (cardiomyocytes) or 72 (hepatocytes) hours of treatment with petroleum substance extracts was performed using the ImageXPress Micro XL high-content imaging system (Molecular Devices) following staining with fluorescent probes Hoechst 33342, Calcein AM, MitoTracker Orange, and AF488-conjugated phalloidin.

TempO-seq assay: Global changes in gene expression patterns in hepatocytes treated with petroleum extracts for 48 hours were analyzed using a targeted RNA sequencing technology, TempO-Seq™ (BioSpyder Technologies, Inc., Carlsbad, CA) measuring a set of genes selected by an effort organized by NIEHS to represent a surrogate whole transcriptome assay, the S1500. Differentially expressed genes were identified using the DESeq2 R package.

Petroleum substances, prototypical high-production volume UVCBs, can be categorized using global similarities in their bioactivity profiles using multi-parametric HCS of iPSC-derived cell types

In combination with transcriptomics analysis, interpretation of these multidimensional data sets is not limited to biological property-derived groupings for regulatory applications, but may eventually also be informative for mechanistic toxicity evaluations

Physico-chemical descriptors and global compositional analysis using Ion Mobility-Mass Spectrometry further improve the confidence in biological data-based groupings

Together, our findings strengthen the argument that a combinatorial approach using quantitative chemical analysis, high-content in vitro screenings, and subsequent computational data integration and visualization possesses the potential to improve chemical-biological read-across applications

MATERIALS & METHODS

IM-MS spectra for each two representative Other Gas Oils (OGO), Straight Run Gas Oils (SRGO), Vacuum & Hydrotreated Gas Oils (VHGO), and Heavy Fuel Oils (HFO) (A). Computational integration of IM-MS data sets in PetroOrg software2 provides unique feature identification and heteroatom class distribution for each sample (B). The upper panel depicts the total number of features for each individual sample. Averages for the top 10 most abundant heteroatom classes are shown in the lower panel.

1Michelmann et al. (2014) J Am Soc Mass Spectrom. 26: 14-24

2www.petroorg.com

3Grimm et al. (2015) Assay Drug Dev Technol. 13: 529-546

4Sirenko et al. (2013) Toxicol Appl Pharmacol. 273: 500-507

5Sirenko et al. (2013) J Biomol Screen. 18: 39-53

6Sirenko et al. (2015) Assay Drug Dev Technol. 12: 43-52

7https://federalregister.gov/a/2015-08529

8Robin et al. (2011) BMC Bioinformatics. 12: 77

9US EPA (2011) Benchmark Dose Technical Guidance

10Reif et al. (2013) Bioinformatics. 29: 402-403

QR-CODE

The authors appreciate useful discussions and technical support from Joel McComb (BioSpyder

Technologies Inc., Carlsbad, CA), Yi-Hui Zhou (North Carolina State University), Grace Chappell (Texas A&M

University), John Braisted, and David Gerhold (National Institutes of Health/ National Center for Advancing

Translational Sciences, Bethesda, MD). Petroleum substances were provided by Concawe, Brussel, BE).

This work was supported by EPA STAR grant #RD83516601 and institutional support from Texas A&M

University. Fabian Grimm is a recipient of SOT Colgate-Palmolive postdoctoral fellowship.

FUTURE WORK

A B

Plots visualize unique feature distributions of the (most abundant) N1 heteroatom class, i.e. carbon chain length vs double bond equivalents (DBE) for each two representative OGOs, SRGOs, VHGOs, and HFOs.

A B C

Heatmap representation (A), cluster analysis (B), and principal components analysis (C) of relative abundances of heteroatom classes using the hclust function in R revealed substance and group-specific fingerprints of petroleum substances.

A B

Extensions of both chemical analysis and in vitro screenings to additional petroleum manufacturing classes Inclusion of a more comprehensive set of tissue types, i.e. iPSC-derived neurons, macrophages, endothelial cells,

and skeletal myoblasts Estimating the interindividual and population variablity in phenotypic responses

A B