CoChat

Kinase inhibitor screening using artificial neural networks and engineered cardiac biowires

Open access

The DOI matches its Crossref record; 64 of the 64 references it cites were found in a registry or an index.

Abstract

Kinase inhibitors are often used as cancer targeting agents for their ability to prevent the activation of cell growth and proliferation signals. Cardiotoxic effects have been identified for some marketed kinase inhibitors that were not detected during clinical trials. We hypothesize that more predictive cardiac functional assessments of kinase inhibitors on human myocardium can be established by combining a high-throughput two-dimensional (2D) screening assay and a high-content three-dimensional (3D) engineered cardiac tissue (Biowire TM) based assay, and using human induced pluripotent stem cell-derived CMs (hiPSC-CMs). A subset (80) of compounds from the GlaxoSmithKline published kinase inhibitor set were tested on hiPSC-CM monolayers and significant effects on cell viability, calcium transients, and contraction frequency were observed. Artificial neural network modelling was then used to analyze the experimental results in an efficient and unbiased manner to select for kinase inhibitors with minimal effects on cell viability and function. Inhibitors of specific interest based on the modeling were evaluated in the 3D Biowire tissues. The three-dimensional Biowire platform eliminated oversensitivity in detecting both Ca 2+ transient amplitude enhancements as well as the acute detrimental effects on cell viability due to the kinase inhibitor application as compared to the monolayer testing.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41598-017-12048-5

Cites

64 of 64 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Kinase inhibitor screening workflow. (A) Cell number, viability, and Ca2+ transients were first assessed in a monolayer format using high-throughput screening assays. (B) Monolayer results were used in an ANN to model and predict the effects of the kinase inhibitors. (C) Inhibitors of interest from the ANN were then tested in 3D Biowire system.
Figure 1 Kinase inhibitor screening workflow. (A) Cell number, viability, and Ca2+ transients were first assessed in a monolayer format using high-throughput screening assays. (B) Monolayer results were used in an ANN to model and predict the effects of the kinase inhibitors. (C) Inhibitors of interest from the ANN were then tested in 3D Biowire system.CC BY · PubMed Central
Figure 2 Validation of monolayer testing assays using control compounds. (A) Representative traces of calcium transient experiments on iPSC-derived CMs after 20-minute exposure to known compounds with negative chronotropic and inotropic properties, nifedipine and thapsigargin. (B) Representative images taken from the cell viability experiments on iPSC-derived CMs after 24-hour exposure to the compounds. (C) Comparison of the effects of control drugs on CMs. Results are represented as the mean value +/− standard deviation. The solid line in each graph represents the results from the blank wells for each parameter, with one standard deviation above and below the average in dotted lines. Results were similar for both nifedipine and thapsigargin; Significant differences from blank controls are indicated by #.
Figure 2 Validation of monolayer testing assays using control compounds. (A) Representative traces of calcium transient experiments on iPSC-derived CMs after 20-minute exposure to known compounds with negative chronotropic and inotropic properties, nifedipine and thapsigargin. (B) Representative images taken from the cell viability experiments on iPSC-derived CMs after 24-hour exposure to the compounds. (C) Comparison of the effects of control drugs on CMs. Results are represented as the mean value +/− standard deviation. The solid line in each graph represents the results from the blank wells for each parameter, with one standard deviation above and below the average in dotted lines. Results were similar for both nifedipine and thapsigargin; Significant differences from blank controls are indicated by #.CC BY · PubMed Central
Figure 3 Effect of kinase inhibitors on hiPSC-derived CM monolayers. Each inhibitor was applied to the cells at 0.1 µM, 1 µM and 10 µM in triplicate. The average of each set of triplicates was compared to the control average, and coded as being within l, 2 or 3 standard deviations of the control compound effect. Heat maps were drawn to visualize changes in normalized cell viability, live cell number, Ca2+ transient magnitude, and Ca2+ transient frequency under spontaneous beating. Inhibitors are listed by experimental number and grouped according to kinase target for clarity. To decode experimental labels, PKIs names are provided in Supplemental Table 1.
Figure 3 Effect of kinase inhibitors on hiPSC-derived CM monolayers. Each inhibitor was applied to the cells at 0.1 µM, 1 µM and 10 µM in triplicate. The average of each set of triplicates was compared to the control average, and coded as being within l, 2 or 3 standard deviations of the control compound effect. Heat maps were drawn to visualize changes in normalized cell viability, live cell number, Ca2+ transient magnitude, and Ca2+ transient frequency under spontaneous beating. Inhibitors are listed by experimental number and grouped according to kinase target for clarity. To decode experimental labels, PKIs names are provided in Supplemental Table 1.CC BY · PubMed Central
Figure 4 Neural network approach to model and predict kinase inhibitors with significant effects on CMs. (A) Several network architectures were examined by altering number of hidden layer(s) and the number of neurons within them, as well as changing the transfer function of each neuron, with the goal of maximizing R2 while minimizing RMSE. The network chosen is highlighted in red. (B) Validation of network performance. After 1000 epochs (iterations), the RMSE converges on a single minimum value, 0.1011. C) Comparison between actual and predicted normalized Ca2+ peak magnitude. There is a correlation between the actual normalized peak magnitude values and the predicted normalized peak values, with an R2 value of 0.9379. The expected trend line is illustrated as the red dotted line. (D) Design parameters chosen for prediction network. These parameters were chosen in order to identify compounds that had the greatest effect on Ca2+ transient magnitude, while minimizing undesired effects on normalized cell viability, live cell number, and frequency. From these limitations, we designed our network with 125,840 data points. The data points used in training the network were normalized as follows: Normalized data = (Experimental data – Blank data)/ Blank data. (E) Predicted normalized value of Ca2+ peak magnitude for each inhibitor. There are 1573 data points for each compound (F) The mean predicted value for each compound is illustrated by the blue line. The compound with the least detrimental effect was found to be inhibitor GW703087X (#44), an EGFR/ErbB2 inhibitor; the compound with the most significant negative effect was found to be inhibitor GW829055X (#24), an GSK3β inhibitor. PKIs names corresponding to the kinase inhibitor numbers shown in E and F are provided in Supplemental Table 1.
Figure 4 Neural network approach to model and predict kinase inhibitors with significant effects on CMs. (A) Several network architectures were examined by altering number of hidden layer(s) and the number of neurons within them, as well as changing the transfer function of each neuron, with the goal of maximizing R2 while minimizing RMSE. The network chosen is highlighted in red. (B) Validation of network performance. After 1000 epochs (iterations), the RMSE converges on a single minimum value, 0.1011. C) Comparison between actual and predicted normalized Ca2+ peak magnitude. There is a correlation between the actual normalized peak magnitude values and the predicted normalized peak values, with an R2 value of 0.9379. The expected trend line is illustrated as the red dotted line. (D) Design parameters chosen for prediction network. These parameters were chosen in order to identify compounds that had the greatest effect on Ca2+ transient magnitude, while minimizing undesired effects on normalized cell viability, live cell number, and frequency. From these limitations, we designed our network with 125,840 data points. The data points used in training the network were normalized as follows: Normalized data = (Experimental data – Blank data)/ Blank data. (E) Predicted normalized value of Ca2+ peak magnitude for each inhibitor. There are 1573 data points for each compound (F) The mean predicted value for each compound is illustrated by the blue line. The compound with the least detrimental effect was found to be inhibitor GW703087X (#44), an EGFR/ErbB2 inhibitor; the compound with the most significant negative effect was found to be inhibitor GW829055X (#24), an GSK3β inhibitor. PKIs names corresponding to the kinase inhibitor numbers shown in E and F are provided in Supplemental Table 1.CC BY · PubMed Central
Figure 5 Viability of Biowire tissues after 24 hr exposure to kinase inhibitor molecules. (A) Representative images of Biowire tissues taken using confocal microscopy after 24 hr exposure to kinase inhibitors. Images represent longitudinal cross-sections of tissues. All cell nuclei were stained with DAPI (blue channel) and dead cells were stained with propidium iodide (red channel). (B) Comparison of the effect of inhibitor GW703087X (#44), GW829055X (#24), and Sunitinib on tissue viability. ANOVA analysis indicated no statistically significant differences in viability between the different drug samples and the control.
Figure 5 Viability of Biowire tissues after 24 hr exposure to kinase inhibitor molecules. (A) Representative images of Biowire tissues taken using confocal microscopy after 24 hr exposure to kinase inhibitors. Images represent longitudinal cross-sections of tissues. All cell nuclei were stained with DAPI (blue channel) and dead cells were stained with propidium iodide (red channel). (B) Comparison of the effect of inhibitor GW703087X (#44), GW829055X (#24), and Sunitinib on tissue viability. ANOVA analysis indicated no statistically significant differences in viability between the different drug samples and the control.CC BY · PubMed Central
Figure 6 Effect of kinase inhibitor drugs on tissue contraction. (A) Active tension, peak duration, contraction slope, and relaxation slope were characterized for each tissue after acute and prolonged exposure to inhibitors GW703087X (#44) (n = 4), GW829055X (#24) (n = 3), and Sunitinib (n = 3). Results are expressed as mean +/− standard deviation. No statistically significant differences were observed for inhibitor GW703087X (#44) at either acute or 24 hr time points. There was a statistically significant decrease in contraction slope and relaxation slope after 24 hr exposure to inhibitor GW829055X (#24), indicated by #, and a significant decrease in relaxation slope between 10 μM acute and long term exposure, indicated by *. An acute effect on active tension, contraction slope, and relaxation slope was observed on tissues treated with the control compound Sunitinib when compared to baseline values, indicated by #. Differences between doses are indicated by *. (B) Representative traces of tissue contraction after acute and prolonged (24 hr) exposure to inhibitor GW703087X (#44), GW829055X (#24), and Sunitinib (stimulated at 1 Hz). Traces were acquired using ImageJ particle tracking software and converted to force using specialized MATALB software.
Figure 6 Effect of kinase inhibitor drugs on tissue contraction. (A) Active tension, peak duration, contraction slope, and relaxation slope were characterized for each tissue after acute and prolonged exposure to inhibitors GW703087X (#44) (n = 4), GW829055X (#24) (n = 3), and Sunitinib (n = 3). Results are expressed as mean +/− standard deviation. No statistically significant differences were observed for inhibitor GW703087X (#44) at either acute or 24 hr time points. There was a statistically significant decrease in contraction slope and relaxation slope after 24 hr exposure to inhibitor GW829055X (#24), indicated by #, and a significant decrease in relaxation slope between 10 μM acute and long term exposure, indicated by *. An acute effect on active tension, contraction slope, and relaxation slope was observed on tissues treated with the control compound Sunitinib when compared to baseline values, indicated by #. Differences between doses are indicated by *. (B) Representative traces of tissue contraction after acute and prolonged (24 hr) exposure to inhibitor GW703087X (#44), GW829055X (#24), and Sunitinib (stimulated at 1 Hz). Traces were acquired using ImageJ particle tracking software and converted to force using specialized MATALB software.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 64 references

Cited by

See all 25

Metadata

Authors
Conant, G., Ahadian, S., Zhao, Y., & Radisic, M.
Affiliations
  • University of Toronto (CA)
  • University Health Network (CA)
Corresponding author
Radisic, Milica
Venue
Scientific Reports
Published
2017-09-18
Volume
7
Issue
1
Pages
11807-11807
ISSN
2045-2322 (ISSN-L)
OpenAlex
W2754401477
Cited by
25
References
64
Citations by year
2026: 1 · 2024: 1 · 2023: 1 · 2022: 2 · 2021: 1 · 2020: 9 · 2019: 5 · 2018: 5
FWCI
4.02
Type
journal article
Language
en
Open access
Licence
Locations
  • Scientific Reports (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
Funders
  • California HIV/AIDS Research Program
  • National Institutes of Health (2r01hl076485-16, 2R01 HL076485, HL076485)
  • Canadian Institutes of Health Research (CHRP-493737-16, MOP-126027 and MOP-137107, 137107_1, 126027_1)
  • Natural Sciences and Engineering Research Council of Canada (CHRP 493737-16, RGPIN 326982‐10, 493737-16)
  • National Heart, Lung, and Blood Institute (R01-HL076485)
Topics
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › Cardiac electrophysiology and arrhythmias
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › Chemotherapy-induced cardiotoxicity and mitigation
  • Health Sciences › Veterinary › Small Animals › Animal testing and alternatives
Keywords
kinase inhibitors, cardiotoxicity, human induced pluripotent stem cell-derived cardiomyocytes, artificial neural networks, high-content screening, calcium transients, cardiomyocyte contraction, drug repurposing, cardioprotection
MeSH terms
Cell Survival, Humans, Myocardium, Neural Networks, Computer, Tissue Engineering, Myocytes, Cardiac, Protein Kinase Inhibitors, Induced Pluripotent Stem Cells

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Kinase inhibitor screening using artificial neural networks and engineered cardiac biowires. https://app.cochat.ai/papers/10.1038/s41598-017-12048-5
The paper
Conant, G., Ahadian, S., Zhao, Y., & Radisic, M. (2017). Kinase inhibitor screening using artificial neural networks and engineered cardiac biowires. Scientific Reports. https://doi.org/10.1038/s41598-017-12048-5

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.