Improved Stem Cell Model Can Test Drugs For Kidney Toxicity

Daniele Zink PhD Institute of Bioengineering and Nanotechnology Singapore

Dr. Zink

MedicalResearch.com Interview with:
Daniele Zink PhD
Institute of Bioengineering and Nanotechnology
Singapore 

Medical Research: What is the background for this study?

Dr. Zink: The kidney is one of the main target organs for toxic effects of drugs, environmental toxicants and other compounds. Renal proximal tubular cells (PTCs) are frequently affected due to their roles in compound transport and metabolism. Validated and accepted assays for the prediction of PTC toxicity in humans currently do not exist. Recently, we have developed the first and only pre-validated assays for the accurate prediction of PTC toxicity in humans 12. This previous work was performed with human primary renal proximal tubular cells (HPTCs) or embryonic stem cell-derived HPTC-like cells. HPTCs are associated with a variety of issues that apply to all kinds of primary cells, such as cell sourcing problems, inter-donor variability and limited proliferative capacity. Embryonic stem cell-derived cells are associated with ethical and legal issues. These are the main reasons why induced pluripotent stem cell (iPSC)-derived cells are currently a favored cell source for in vitro toxicology and other applications.

The problem was that stem cell-based approaches were not well-established with respect to the kidney. Recently, the group of IBN Executive Director Prof. Jackie Y. Ying developed the first protocol for differentiating embryonic stem cells into HPTC-like cells, and my group has contributed to characterizing these cells and publishing the results 3.  In the work published in Scientific Reports ,4we have applied a modified version of this protocol to iPSCs. In this way, we have established the simplest and fastest protocol ever for differentiating iPSCs into HPTC-like cells. The cells can be used for downstream applications after just 8 days of differentiation. These cells can also be applied directly without further purification due to their high purity of > 90%.

By using these cells, we have developed the first and only iPSC-based model for the prediction of PTC toxicity in humans. This was achieved by combining our iPSC-based differentiation protocol with our previously developed assay based on interleukin (IL)6/IL8 induction 12 and machine learning methods 5. Machine learning methods were used for data analysis and for determining the predictive performance of the assay. The test accuracy of the predictive iPSC-based model is 87%, and the assay is suitable for correctly identifying injury mechanisms and compound-induced cellular pathways.

Medical Research: What are the main findings?
Dr. Zink:

– Established the fastest and simplest protocol for differentiating iPSCs into HPTC-like cells (1 step, 8 days). The fastest alternative protocol 6 comprises 4 steps and requires at least 11 days.

– The iPSC-derived HPTC-like cells obtained with our protocol are highly pure (>90%) and can be directly used for subsequent applications. The purity of cells derived from alternative protocols has not been determined and any applications based on such cells have not yet been developed.

– Established the first and only iPSC-based model for the accurate prediction of PTC toxicity in humans.

–  The model combines iPSC technology with machine learning methods for automated and unbiased data analysis. Data analysis by machine learning also improves the accuracy of the model, and the current test accuracy is 87%. Data analysis and machine learning was done by the group of Dr. Lit-Hsin Loo at the Bioinformatics Institute (A*STAR).

– Apart from being highly predictive, the model also identifies compound-induced injury mechanisms and cellular pathways correctly.

Medical Research: What should clinicians and patients take away from your report?

Dr. Zink: In the future, our technology could help to develop safer drugs, consumer products and other compounds (e.g. industrial chemicals).

Nephrotoxicity is typically detected late in drug development, accounting for 2% of drug attrition during pre-clinical studies and 19% in phase 3. In some cases, nephrotoxicity is detected only after the drug is launched for example in the case of tenofovir (toxic for PTC in humans). Late detection of nephrotoxicity may be harmful to the patients, and incurs high costs for the pharmaceutical industry.  Our iPSC-based model could help to predict PTC toxicity with high accuracy during the early pre-clinical stages. This would reduce costs for the pharmaceutical industry and help to develop safer drugs.

In fact, many marketed drugs are nephrotoxic, and due to the problems outlined above it is difficult to develop less nephrotoxic drugs. This causes many problems for clinicians (for a more detailed discussion see 7). For instance, immunosuppressive drugs like tacrolimus and cyclosporine A are already nephrotoxic at standard recommended doses. Such drugs must be taken by kidney transplant patients to suppress rejection. If the dose is too low, the transplanted organ gets rejected. If the dose is too high the transplanted organ gets destroyed by the toxicity of the drugs. The dilemma for clinicians is to find the right individual dose amid the uncertainty of whether episodes of compromised kidney function are due to dosage that is too high or too low.

Nephrotoxic drugs are a major cause why 5-7% of hospitalized patients develop acute kidney injury (AKI), and the incidence in intensive care units is 30-60%. There are often no alternatives to nephrotoxic drugs for the treatment of critically ill patients. For instance, the nephrotoxic antibiotic gentamicin is widely used for the treatment of sepsis. In critically ill patients, AKI that requires dialysis is associated with a mortality of 40-70%, and AKI is an independent predictor of death (for references see 7).

In addition, our model would be relevant for the following industries and customers/consumers:

Cosmetics industry: There is strong pressure to use alternative methods due to the animal ban in the EU (7th Amendment of the EU Cosmetics Directive 76/768/EEC). Similar laws have been implemented in Israel and India, while China and Korea are expected to follow.

Chemicals industry: The production of large, increasing numbers of compounds with often unknown toxicity requires novel efficient and predictive screening systems. This largely unmet need is reflected by new programs in the US (ToxCast and Tox21), and further enhanced by altered legislation (REACH in the EU).

Food and nutrition and consumer care industries: These industries require efficient, predictive test systems for the safety evaluation of novel ingredients and compounds.

Medical Research: What recommendations do you have for future research as a result of this study?

Dr. Zink:

–       We have recently developed a method that is suitable for high-content screening (HCS). HCS is the industrial standard for screening of large compound libraries. The HCS method predicts nephrotoxicity in humans with high accuracy. We are publishing a paper on this work, which was done in collaboration with Dr. Lit-Hsin Loo’s group from the Bioinformatics Institute, who also developed the machine learning methods described in our recent publication in Scientific Reports.

–       We also work on predictive microfluidic models for repeated dose and chronic toxicity testing.

Citation: 

Kandasamy K, Chuah JK, Su R, et al. Prediction of drug-induced nephrotoxicity and injury mechanisms with human induced pluripotent stem cell-derived cells and machine learning methods. Sci Rep. 2015;5:12337.

References:

  1. Li Y, Kandasamy K, Chuah JKC, Lam YN, Toh WS, Oo ZY, Zink D: Identification of nephrotoxic compounds with embryonic stem cell-derived human renal proximal tubular-like cells. Molecular Pharmaceutics,111982-1990, 2014
  2. Li Y, Oo ZY, Chang SY, Huang P, Eng KG, Zeng JL, Kaestli AJ, Gopalan B, Kandasamy K, Tasnim F, Zink D: An in vitro method for the prediction of renal proximal tubular toxicity in humans. Toxicol Res,2352-362, 2013
  3. Narayanan K, Schumacher KM, Tasnim F, Kandasamy K, Schumacher A, Ni M, Gao S, Gopalan B, Zink D, Ying JY: Human embryonic stem cells differentiate into functional renal proximal tubular-like cells. Kidney Int,83593-603, 2013
  4. Kandasamy K, Chuah JK, Su R, Huang P, Eng KG, Xiong S, Li Y, Chia CS, Loo LH, Zink D: Prediction of drug-induced nephrotoxicity and injury mechanisms with human induced pluripotent stem cell-derived cells and machine learning methods. Sci Rep,512337, 2015
  5. Su R, Li Y, Zink D, Loo LH: Supervised prdiction of drug-induced nephrotoxicity based on interleukin-6 and -8 expression levels. BMC Bioinformatics,15(Suppl. 16)S16, in press
  6. Lam AQ, Freedman BS, Morizane R, Lerou PH, Valerius MT, Bonventre JV: Rapid and efficient differentiation of human pluripotent stem cells into intermediate mesoderm that forms tubules expressing kidney proximal tubular markers. J Am Soc Nephrol,251211-1225, 2014
  7. Tiong HY, Huang P, Xiong S, Li Y, Vathsala A, Zink D: Drug-Induced Nephrotoxicity: Clinical Impact and Preclinical in Vitro Models. Mol Pharm,111933-1948, 2014

[wysija_form id=”5″]

 

Daniele Zink PhD (2015). Improved Stem Cell Model Can Test Drugs For Kidney Toxicity 

Last Updated on October 28, 2015 by Marie Benz MD FAAD