Generating iPSCs by various reprogramming methods

Gene signatures of human induced pluripotent stem cells may uncover which reprogramming method was used

Go to the profile of Jared Churko
Oct 08, 2017
Upvote 0 Comment

When I started my postdoctoral fellowship with Joseph Wu at Stanford University, the methods used to generate human induced pluripotent stem cells (hiPSCs) were primarily lentiviral-based approaches using fibroblasts as a cell source. At this time, an international effort began to generate hiPSCs from various methods as well as from various cell sources. Our lab had initially generated a minicircle method to generate hiPSCs, but we were also working on an improved minicircle construct. At the same time, a Sendai-virus-based method also became available along with an mRNA and microRNA/mRNA based method. We were utilizing each of these reprogramming methods in the lab and wondered which method makes ‘better’ hiPSCs. To address this question, we compared it to the gold standard of human embryonic stem cells (hESCs) to find out whether a certain reprogramming method generated hiPSCs more similar to hESCs. By taking a whole transcriptome (RNA-seq) and epigenome (ChIP-seq) approach, we investigated the subtle differences between hiPSCs and hESCs. The results are reported in our paper recently published in Nature Biomedical Engineering.

The greatest use of hiPSCs is that they can be differentiated into various cell types. Our lab focuses on using hiPSCs to study heart diseases and we routinely generate cardiomyocytes from hiPSCs. Hence, we hypothesized that if hiPSCs from a specific reprogramming method could terminally differentiate into cardiomyocytes to a greater extent, this method may be superior to other reprogramming methods. However, each reprogramming method varies in its reprogramming efficiency, cost, safety concerns, and technical difficulty. Furthermore, due to the cost and technical difficulties in deriving multiple hiPSC lines, attempts to compare hiPSCs have required analysis of hiPSCs generated in multiple laboratories. During this time, our lab was also part of the progenitor cell biology consortium (PCBC) which created the Cincinnati Cell Characterization Core, with a primary goal of comparing hiPSCs generated by multiple methods from various institutions. When assessing transcriptomic difference from cell lines generated from multiple labs, it was reported that there is a “strong correlation between gene expression signatures and specific laboratories in both ESC and iPSC lines”1. This suggests that hiPSCs created from different labs, different reprogramming method, different genetic backgrounds, and different culturing conditions can impact the comparative analysis being performed. However, to accurately compare differences only attributed to the reprogramming method, we needed to ensure all hiPSCs lines were generated under standardized culture conditions (same seeding density, passaged equally, same enzymatic method of passaging (Accutase) and same growth media (E8 media)). In addition, the genetic background of the hiPSCs play a significant role in their underlying gene expression profile so we wanted to use the same female fibroblast population (equal genetic background) for all of our hiPSC lines. Given that all of our hiPSC lines were from a female donor, all of the hESCs that we used for comparison were also female. Lastly, we wanted to be as comprehensive as possible in our reprogramming methods, so we chose six most widely used reprogramming methods (lentivirus, episomal, minicricle, mRNA, microRNA, and Sendai viral based methods). However, due to the then high cost (sequencing costs) and technical difficulty (some methods took a lot longer to generate hiPSC lines) in generating these lines using all six reprogramming methods, we were able to study one donor line (Figure 1). 

Figure 1. Standardized comparison of various reprogramming methods. The same female fibroblast line was reprogramed using six common reprogramming methods (lentivirus, episomal, minicricle, mRNA, microRNA, and Sendai virus). At passage 4 (p4) lines were frozen until all lines were made. Twelve hiPSC lines were equally passaged using accutase and cultured on matrigel with E8 media. At passage 12 (p12), all lines were assessed by RNA-seq and ChIP-seq.

To comprehensively compare these lines, we performed high coverage RNA-seq on total RNA. While clones generated from different reprogramming methods were highly similar, clones generated by the same reprogramming method clustered together suggesting that the reprogramming method contributes to the gene differences reported when comparing hiPSCs to hESCs. We went on to observe that a major difference between hiPSCs and hESCs is the activity of the polycomb repressive complex (PRC). In addition, anecdotal evidence from our lab suggested that some hiPSC lines differentiated more efficiently. To determine if this could be due to the reprogramming method, passage number, genetic background or to differences during each differentiation batch (seeding density, timing of small molecule, small molecule batch differences), we standardized all of these conditions and performed cardiomyocyte differentiation from all lines in parallel. Somewhat not surprisingly, we found that the reprogramming method chosen does not greatly influence the differentiation potential. However, the differences in differentiation potentials observed may still be due to the genetic background of the patient or passage number. This is also echoed in our assessment of XIST expression demonstrating that XIST levels decrease over prolonged passage (p12 versus p40) and suggest that even at passage 12, hiPSC lines are still undergoing transcriptomic and epigenetic changes. However, observations not included in our manuscript suggest an increase in genetic abnormalities (in using SNP arrays) when assessing p40 hiPSCs.

In summary, while many reprogramming methods were utilized in our study, the majority of our hiPSC lines are now generated using Sendai virus. This is due to the ease in generating hiPSCs using this method as well as the little to no chance of transgene integration. Nevertheless, it is interesting to note that while this method remains the most popular reprogramming method, in our study the two Sendai-virus-derived lines did not cluster as close together as clone pairs from other methods. We went on to observe by qPCR that persistent Sendai virus was expressed within one of our lines even at passage 12. We have also observed in some of our in house lines demonstrated persistent Sendai virus up till passage 22. This suggests that regardless of passage number, cell source, or reprogramming method utilized, each hiPSC line still requires robust characterization prior to them being used for downstream experimentation or clinical use. 

Our paper: Churko, J. M. et al. Nat. Biomed. Eng. (2017) doi:10.1038/s41551-017-0141-6.

References: Newman A. M. & Cooper, J. B. Cell Stem Cell. 7, 258-262 (2010) doi:10.1016/j.stem.2010.06.016.

Go to the profile of Jared Churko

Jared Churko

Instructor, Stanford

No comments yet.