Review of Important New Yamanaka Paper on Defective iPS cells

A team led by Shinya Yamanaka has published a new study reporting analysis of defective induced pluripotent stem (iPS) cells in the journal PNAS.

Part of what I liked so much about this paper is that it tackled a very real, but controversial area of stem cell research. By this I mean the reality that a significant number of iPS cell lines are not always entirely identical to embryonic stem cells and some, a minority, have important defects. My own lab reported on another part of this reality with our paper late last year indicating that reprogramming has a significant number of features in common to oncogenic transformation.

Yamanaka Paper

In the new paper, Koyanagi-Aoi, et al., the team reports that some human iPS cell (hIPSC) clones that by other criteria seem to be “true” iPS cells are indeed abnormal as judged by their failure to differentiate normally.

Although hESC and hIPSCs are remarkably similar on many levels overall as indicated both in this paper and in much previous research, Yamanaka’s team report here that a significant number of hIPSC lines not only behaved differently, but also had expression changes in specific genes. These changes could be important for explaining why these hIPSC lines did not behave properly.Interestingly, the implicated genes included expressed endogenous retroviruses and those with specific associated elements called long terminal repeats (LTRs) and in particular one called LTR7.

Table 1 of the paper really tells a lot of the story as it lists the cells and methods used to make the hIPSC clones and which ones were defective. The authors break down the abnormal clones into “type-1” and “type-2”. Type-1 refers to lines that retained OCT3/4+ undifferentiated cells in the differentiated culture prior to transplantation. Type-2 refers to those lines that failed to differentiate properly but did not show high expression of OCT3/4 prior to transplant.

They mention that a total of 49 hIPSC lines were analyzed, of which a quite a lot had either type-1 or type-2 problems. In fact, of 63 total grafts, 22% exhibited type-2 defects. A significant, but smaller percentage had type-1 defects.

Remarkably, even hIPSC lines generated with episomal vectors showed significant rates of these defects. In contrast, encouragingly Sendaivirus methods produced no defective clones, although only 18 clones were in total generated using this approach.

The defects they did observe in some lines certainly have clinical implications because when the team transplanted neural derivatives of the defective hIPSC lines into mouse brains, they formed teratoma. Thus, even after differentiation, these hIPSC lines could not be considered safe for clinical use. This raises the broader question of how we can be sure that any particular hIPSC line is clinical grade? Clearly significant molecular and functional analysis of hIPSC lines is crucial prior to contemplating their even indirect clinical use via differentiated derivatives.

So what can we look for to distinguish good from bad or defective hIPSC clones?

Many of the standard hIPSC validation assays done in Yamanaka’s paper (and others) should also be on the check list for any hIPSC lines potentially intended to be used clinically in any way. In addition, the team implicates a key suspect in the form of endogenous retroviruses containing LTR7 elements and associated genes.

Figure 3B is shown above, indicating how some LTR7-related genes are overexpressed in defective hIPSC clones. They also found that some of the defective hIPSC clones exhibited abnormal hypomethylation of LTR7 as well. Again, these clones were predisposed, even after differentiation, to form teratoma after transplantation in mouse brains.

Interestingly, the hIPSC-related defects were found in zero out of 10 hESC lines. This might suggest again how in some ways hESC have specific differences that make them more suitable for certain clinical applications. More than 10 hESC lines need to be analyzed of course.

More studies will be needed as well on the hIPSC front because of the ways in which specific founder cells tended to be made into hIPSC using specific methods. In this regard, the authors say:

Future studies will need to be undertaken to determine whether the origin or the generation method (or both) has a significant impact on the frequency of type-1 differentiation-defective iPSCs.

The authors sum up as follows:

In conclusion, our results revealed that a subset of hiPSCs is

defective in neural differentiation and marked with activation of endogenous retroviruses. We also confirmed that some hiPSCs are different from hESCs in molecular signatures, including CGDMRs, which has been previously reported. It remains to be determined whether these molecular signatures specific for some hiPSCs have functional consequences.

Overall, I found this paper overall to be excellent and intriguing.

I don’t find it discouraging because frankly in the field we knew this kind of thing was the reality, now we have more information about how to pick the best hIPSC lines for clinical use, and most hIPSC lines are in fact not defective.


  1. I particularly like this paper because it provides the first really concrete criteria for predicting the tumorigenicity of pluripotent cells, which is a big step. Lots more work to do on this front to ensure we can make safe cells for transplantation…

  2. I liked this paper, too, for a couple of reasons. First, it drives home the message that all hESC and hiPSC derivatives that are destined for clinical use MUST be analyzed thoroughly, not just by immunocytochemistry or karyotyping, but by genomic and epigenetic methods. These methods are actually quite inexpensive, especially when compared to the costs of giving a patient abnormal cells. As you know, we already have a ton of gene expression, SNP genotype, and DNA methylation data on undifferentiated and differentiated hESCs and hiPSCs available for anyone to use for comparison, with whole genome sequencing coming soon.
    Second, we have been using Sendai for reprogramming for years, and wondered if we should consider episomal vectors. This paper suggests that Sendai might be better, or at least no worse.

    • Thanks, Jeanne, for your comment and insight.

      It seems to me that an additional vital test, not so affordable, is function: transplantation of derivatives of hIPSCs in animal models in the target organ.

      Also, what do you think of the endogenous retrovirus connection?

      • Hi Paul:
        I’m just starting to consider the endogenous viruses…first I want to map them- whole genome sequencing tosses the sequences because they aren’t in the reference genome. The animal transplantation will need to be done early in the approval process, and I would hope that the analysis would include looking at differentiation of the cells.

5 Trackbacks / Pingbacks

  1. Review of Important New Yamanaka Paper on Defective iPS cells
  2. iPS Method/Team Update on Path to Clinic - Page 2
  3. News and Blog Roundup – 22/11/13 | Stu's Stem Cell Blog
  4. Cells Weekly – November 24, 2013 | Stem Cell Assays
  5. Review of Important New Yamanaka Paper on Defective iPS cells | TransBio-Tex

Comments are closed.