Review of Mitalipov group Nature paper: cloned ES cells versus iPS cells

Just how good are human embryonic stem (ES) cells made by therapeutic cloning via nuclear transfer?

How do they compare to induced pluripotent stem (iPS) cells or traditional ES cells made from IVF embryos?

A new paper in Nature directly tackles these key questions, but first a bit of context.

Three separate groups have now successfully made ES cells using somatic cell nuclear transfer (SCNT), with the successful technique first reported by the lab of Shoukhrat Mitalipov at OHSU last year.

I have reviewed those three therapeutic cloning papers (see here, here and here).

It’s worth noting that I felt that therapeutic human cloning was the top stem cell story of 2013. For an illustration of how therapeutic cloning differs from reproductive cloning see a handy illustration here.

It is pretty clear then that transferring a somatic cell nucleus into an oocyte in place of its own nucleus is a new, now established method of making human ES cells and this new kind of pluripotent stem cells (termed NT ES cells) is yet another potential tool for stem cell-based regenerative medicine.

However, let’s return to the big question: given the power, flexibility, and relative ease of making iPS cells, is there good reason to go to all that trouble to make NT ES cells?

Or to put it more bluntly, are NT ES cells any better than iPS cells in some important way to justify how much more of a pain they are to make?

Mitalipov’s lab has a new Nature paper (Ma, et al.) that argues strongly “yes” to that question. The paper is entitled “Abnormalities in human pluripotent cells due to reprogramming mechanisms”.

In this new paper published today, Mitalipov’s group teamed up with several others has done a comprehensive genomic, epigenetic, and transcriptomic analysis of NT ES cells. They compared them to iPS cells and IVF ES cells with each also compared to parental human dermal fibroblasts (HDFs).

The paper argues that in pretty much every way, NT ES cells are significantly more similar to IVF ES cells than iPS cells were. In other words, the authors argue that if IVF ES cells are the gold standard, NT ES cells are closer to that gold standard than iPS cells are.

NT ESC gene expression

This team also reports that iPS cells have more genomic abnormalities and DNA methylation differences.

They also assert that transcription factor-based reprogramming is inherently just not as good as SCNT at reprogramming. For example, in terms of global gene expression, NT ESCs clustered with IVF ES cells (see Figure 6A above; green and red columns of names across the top, respectively), while iPS cells (orange names) were further way and hence less similar. It’s notable that iPS cells uniquely had abnormal expression of Zinc finger protein coding genes (yellowish region of bars constituting “Cluster 10″ in the bottom right of the heat map).

Overall, the paper is very convincing, but there are a few issues and limitations here that are notable.

First, the paper does not provide mechanistic insight into reprogramming. It is basically saying that the inherent physical nature of NT ES cells is better than that of iPS cells. I would have loved some specific insight into why NT is better than transcription-based reprogramming, but I understand that that is a tough nut to crack.

Second, the paper did not address important cellular functions like differentiation or tumorigenicity. In other words, do the reported differences between NT ES cells and iPS cells actually have any functional consequence? Are they meaningful? The paper does not address the function of the cells. I was left wondering, “How do these cells behave? Are there meaningful functional differences in differentiation or tumorigenicity?”

By analogy you might say you are comparing two boxers who are going to be in a fight soon. You examine and compare their heights, weights, ages, BMIs, and such physical traits, but until you see how they actually box, can you really predict who will do better? Function over form.

To extend this analogy a step further, one boxer (iPS cells) has already boxed in a number of matches over the years and has a proven track record of doing well, while the other (NT ES cells) has never boxed before. I’m sure we’ll learn more from this and other groups about NT ES cell function in the coming months/years.

Third (and related to above), the NT ESC paper does not address the endogenous retroviral elements that Yamanaka has linked to differentiation-defective phenotypes of reprogrammed cells. In their Koyanagi-Aoi, et al. paper (reviewed here), Yamanaka’s team made a strong case that specific retroviral elements should be examined as indicators of defective reprogrammed cells. Why not check out those elements in the NT ESCs?

I am also puzzled at the title of the paper. Why focus on the negative?

A more minor point is that I would have liked to have seen how iPS cells made with episomal vectors compared.

Overall, despite a few issues, I would say this is a very important, thorough, compelling paper.

I still am concerned from a broader perspective about unintentional enabling of human reproductive cloning as well and hope that more dialogue will emerge on that important element.

While from a big picture standpoint I’m still not entirely convinced that NT ES cells (given the egg donation, expense, and complications of making them) will give iPS cells a practical run for the money when it comes to regenerative medicine, at the same time this paper made the case for NT ESCs relatively much stronger.

Cool paper by Cowley Lab on the role of HDACs in ES cells

Cowley HDAC knockoutIt was great being a postdoc in Bob Eisenman’s Lab at the Hutch in Seattle. I loved it.  Bob is a great mentor (see more on my experience there here) and the scientific interactions in the lab were wonderful.

It’s been fun and interesting to follow the work of my fellow former Eisenman lab members over the years. For example, another former postdoc from the Eisenman lab, Shaun Cowley, has a lab that is doing some exciting research on the role of histone deacetylase (HDACs) in stem cells including embryonic stem (ES) cells.

Shaun’s lab has just come out with a great new paper in PNAS on the role of HDAC1 and HDAC2 in ES cells. The paper, Jamaladdin, et al., focuses on using knockouts of HDACs 1 and 2 in ES cells to, in a very elegant way, decipher their functions related to pluripotency and also the molecular machinery that controls HDAC activity.

HDACs 1 and 2 are major regulators of global histone acetylation levels and transcriptome activity in the epigenome. They are recruited at least in part via a corepressor protein of great interest called Sin3. The Mxd/Mad proteins that in part act as Myc antagonists recruit HDACs via Sin3.

Shaun’s lab reported that knockout of HDAC1/2 reduces most of the overall HDAC activity in ES cells, which end up losing viability too. The HDAC deficient cells also do not segregate their chromosomes properly (see Figure 2C above). The HDAC knockout ES cells exhibited some interesting changes in gene expression as well, including reduced expression of key pluripotency factors such as Oct4 and Nanog.

Digging into mechanisms they found that a molecule called inositol tetraphosphate (IP4) controls HDAC activity. While introduction of WT HDAC1 can rescue cell viability in HDAC1/2 knockout ES cells, mutant forms of HDAC1 that have impaired ability to bind IP4 also in turn have substantially reduced ability to rescue HDAC null ES cell viability.

They conclude with some indication of how the research may have clinical impact:

This leads us to predict that specific inhibitors of HDAC1/2 should exhibit selective toxicity toward immortalized cell types, making them effective therapeutic targets in the treatment of cancer.

I enjoyed this paper and recommend it.

Adult human therapeutic cloning of embryonic stem cells by SCNT

An international team of stem cell scientists has replicated human therapeutic cloning to make embryonic stem cells via somatic cell nuclear transfer (SCNT).

The team was led by Drs. Dong Ryul Lee of CHA Stem Cell Institute in Korea and Robert Lanza of Advanced Cell Technology (ACT) and reported the advance in the Chung, et al. paper today in the journal Cell Stem Cell entitled “Human Somatic Cell Nuclear Transfer Using Adult Cells”. The cells expressed pluripotency markers (see Figure 1A at left) and had normal karyotypes.

Human SCNT

The research has replicated the human therapeutic cloning work reported last year by Mitalipov’s group and has advanced the field’s knowledge further in some ways.

One major important element of the new paper is the successful use of adult and elderly somatic cell nuclear donors (ages 35 and 75). Therefore, this new work indicates that SCNT may become a viable option for production of ES cells from in principle almost any person. Further, the new paper suggests that a slightly longer period of incubation prior to activation following SCNT may yield better results.

Some questions and challenges remain, which is not surprising for such a new technology. For example, why is it still relatively speaking so difficult to make SCNT work to make NT-ESCs in humans compared to other animals? An additional hurdle is that efficiency remains a challenge. In the current paper, 2 lines were made from a total of 77 human oocytes.

Again, it is early days for human SCNT and we can be almost certain that further refinements to the technology will boost efficiency.

For perspective, production of iPS cells is also an inefficient process, but the key difference here is that to make human iPS cells one can easily start with an almost unlimited numbers (easily in the 10s of millions) of say skin cells. In contrast, the low efficiency of human SCNT is much more of a challenge because every attempt involves a unique, difficult to obtain human oocyte. In specific countries and states in the US, oocyte procurement faces complicated regulatory and legal hurdles. In this regard, it is worth noting that Mitalipov’s group has just recently reported in Nature successfully conducted mouse SCNT to make ES cell lines using 2-cell embryo cells generated from fertilized eggs rather than naive oocytes. If this works in the human context, egg procurement may become less of a challenging issue.

As with the Mitalipov group paper, this team also did not mention the broader context whereby human SCNT technology could be misused by rogue scientists to pursue human reproductive cloning. As with many powerful technological advances, dual use ethical issues can arise and that is certainly the case here with therapeutic human cloning.

Some interesting questions that remain open include why this SCNT process is relatively inefficient in humans versus other species and also why certain human donors produce eggs that have the “right stuff” for successful SCNT.

Overall, I think this paper is an exciting, important, and technically convincing. SCNT ES cells may give us another potential tool to help patients via stem cells. Mouse studies hint that SCNT ES cells may have some advantages over iPS cells, but the jury is still out on that.


Review of New ACT Paper on hESC-derived MSCs

Advanced Cell Technology (ACT; $ACTC) has a new paper out on using human embryonic stem cells (hESC) to make adult stem cells with potentially powerful therapeutic potential.

The paper, entitled Mesenchymal stem cell population derived from human pluripotent stem cells displays potent immunomodulatory and therapeutic properties, was published in the journal Stem Cells and Development.

What’s the scoop on this paper and this area of adult stem cell preclinical/clinical research?

It’s an interesting, notable paper. At the same time there are some areas in the paper that could have been stronger and the path to an approved, commercially viable hESC-MSC product could be long and challenging with fierce competition.

What’s this arena like for background?

One of the best things about adult stem cells is that we all already have them in our bodies.

For example, there are relatively abundant populations of mesenchymal stem cells–now more becoming more commonly referred to as mesenchymal stromal cells (MSCs)–in both adipose tissue and bone marrow in adults as well as in umbilical cord blood.

Another beneficial aspect to adult stem cells such as MSCs is that they have a relatively very low (although not absent) tumorigenic capability.

These factors make adult stem cells a readily accessible source of material for cellular medicine therapies. Thousands of clinical trials all around the world are already underway using adult stem cells including hundreds on MSCs.

When I first heard generally about the line of research that ended up in this Kimbrel, et al. ACT paper on hESC-derived MSCs last year I have to admit my first reaction was to be a bit puzzled and can be summed up as follows as a question:

why use hESC to make MSCs for potential allogeneic use if we can just relatively easily harvest endogenous MSCs from patients for autologous use?

This question was still on my mind as I read the actual paper.

The authors make several arguments as to why hESC-MSCs are needed, but sum up the case this way:

“Our data suggest that this novel and therapeutically active population of MSCs could overcome many of the obstacles that plague the use of MSCs in regenerative medicine and serve as a scalable alternative to current MSC sources.”

What more specifically are their arguments for hESC-MSCs?

  • More is better. The team can make nearly an unlimited supply of hESC-MSCs, whereas endogenous MSCs have limited proliferative capacity in culture.
  • Higher consistency. The authors argue that hESC-MSCs are more homogeneous and consistent in nature than purified endogenous MSCs or MSCs amplified in culture.
  • Younger is superior. They also assert that hESC-MSCs may for many older/sick patients essentially be younger, healthier cellular versions of their own endogenous MSCs.

As the authors point out, they are not the first to make hESC-MSCs, but they seem to have done it arguably in a simpler, more scalable, and more clinically relevant manner plus investigated the hESC-MSC properties more thoroughly than others have in the past. For example, there is some intriguing data here on the immunomodulatory properties of hESC-MSCs in rodent disease models.

There are also some gaps in the paper.

For instance, while they report that their data indicates that hESC-MSCs have a 30,000-fold greater expansion capability compared to endogenous MSCs (see Figure 2A below), this is a double-edged sword on the safety front.

Endogenous cultured MSCs have a very lower spontaneous immortalization rate, which is a great thing, but it is not zero. Immortalization is the first step toward becoming a cancer cell. When MSCs are grown beyond about 15-20 passages or about 4-5 weeks, there have been enough population doublings and sufficient time that has passed that the relative risk of immortalization and mutations goes way up.


What this means is that while being able to grow hESC-MSCs to and even beyond 30 passages gives a team the ability to make a heckuva lot of potential therapeutic doses of MSCs (which is great), there could well be a direct inverse tradeoff with decreasing safety occurring at the same time.

See how the green and purple curves are beginning to level off after 1 month? That’s the normal lifespan of MSCs in vitro and is protective against tumorigenesis. The relatively straight blue and red more sharply sloped lines means the hESC-MSCs are continuing to grow like gangbusters, but again that growth could come with a safety tradeoff. Or maybe not. Without more data we just don’t know.

More broadly the area of safety of hESC-MSCs was one major area in which I thought the paper could have been much more thorough and needed more depth. In fact, words such as “teratoma”, “immortalization”, “safety”, “tumorigenesis”, and “karyotype” are not used even once in the paper that I could find.

Perhaps the team of authors knows these cells to be safe in a preclinical rodent setting, but it would have been great if that data were included. Perhaps in a 2nd paper?

A second key issue is immunogenicity. The authors assert rather absolutely that hESC-MSCs could be used in an allogeneic manner in human patients without immunosuppression and without negative consequences. I’m not so sure that we know that to be so clearly the case. While hESC-MSCs may have some level of immunoprivilege, immunosuppression could well prove necessary when used in an allogeneic manner. Again this is an issue where at this point we just don’t know. In contrast, even lab-expanded endogenous MSCs when used autologously would not require immunosuppression.

For more helpful background see an excellent recent review in Nature Biotechnology by Jeffrey Karp’s team: Mescenchymal stem cells: immune evasive, not immune privileged. It discusses in a very clear and insightful way the key challenges facing autologous and allogeneic MSC therapies and makes the argument that MSCs are not immunoprivileged. It concludes this way:

“To maximize patient benefit and minimize patient risk, next-generation MSC therapies should be built on a foundation of thorough characterization and fine-tuning of MSC immunogenicity, survival, potency and disease-specific mechanisms of action.”

A third significant issue for the hESC-MSCs is regulatory in nature. Endogenous MSCs, if less than minimally manipulated, are not regulated as biological drugs, while hESC-MSCs are certainly biological drugs and hence subject to far more lengthy and expensive FDA vetting. Of course even endogenous MSCs are often grown in culture as well (and therefore are in that form biological drugs too) and stem cell clinic operators have told me privately that they believe that amplification of MSCs is needed to make an effective product. Even so, I predict that the FDA would view hESC-MSCs as relatively of greater safety risk to patients than amplified endogenous MSCs.

The bottom line is that this ACT hESC-MSC paper is solid and interesting, but just the beginning of the story on ACT hESC-MSCs. There’s a lot more we need to know to judge their clinical potential as well as their relative utility compared to endogenous MSCs, and foremost on my mind is safety. Data on the karyotypes, immortalization rate, and tumorigenicity of the hESC-MSCs grown for months in culture would go a long way toward clearing things up.

Any hESC-MSC-based commercial product would face stiff competition from the numerous commercial entities already years ahead and collectively conducting about 350 clinical trials using either allogeneic or autologous MSC therapies. This doesn’t mean that ACT shouldn’t continue this work, but there are numerous challenges.

Disclosure: I do not currently own stock in ACTC or any competing company.

Cool Cell Stem Cell Paper From Ishii Team on Histones & Reprogramming

ShinagawaTwo of my favorite things are stem cells and histones.

When they both come together in one paper and it’s in my favorite journal, Cell Stem Cell, you’ve got my full attention. My own lab is especially interested in the roles of histones and Myc in stem cells.

So when I saw a paper a paper from a team lead by Shunsuke Ishii at RIKEN on the roles of histones TH2A and TH2B (variants of H2A and H2B), I thought it was very cool.

The paper from Ishii’s team, Shinagawa, et al., is entitled Histone Variants Enriched in Oocytes Enhance Reprogramming to Induced Pluripotent Stem Cells.

They report that the histone variants TH2A and TH2B, which are highly expressed in testes, oocytes, and one-celled embryos, substantially boost the cellular reprogramming process to make iPS cells. There was almost no expression of the variants in MEFs (no surprise) but interestingly there was little expression of the variants in ES cells either. That’s more surprising.

A lot of factors can enhance reprogramming so why is this work exciting? What is particularly interesting here is that it seems that TH2A and TH2B function in a novel kind of way by globally opening chromatin structure (Figure 7). This may set the table for subsequent activity by other reprogramming factors.

Surprisingly, this new paper reports that TH2A and TH2B had no effect on the heterochromatic mark H3K9me3 so these variants must promote open chromatin through a mechanism not involving repression of heterochromatin that is enriched for H3K9me3. For comparison, Myc, which can also boost global chromatin openness at least in part by recruiting histone acetyltransferases, seems to inversely correlate with H3K9me3 based on Myc knockout studies.

The team did link up these histone variants to other epigenetic marks (emphasis mine) in ways that are fascinating even if still not entirely clear mechanistically:

A lack of maternal TH2A/TH2B disturbed activation of the paternal genome in zygotes. In zygotes lacking maternal TH2A/ TH2B, the levels of H3K4me3 on the paternal genome were much lower than those in the control WT zygotes. The levels of 5mC on the paternal genome in the zygotes lacking maternal TH2A/TH2B were higher than those in WT zygotes. In addition, the paternal genome in the zygotes lacking maternal TH2A/TH2B contained lower levels of 5hmC. These results suggest that in the absence of maternal TH2A/TH2B, paternal genome activation, which accompanies H3K4me3 and DNA demethylation, is defective. Consistent with these observations, the levels of H3K27me3 on the paternal genome were higher at the PN5 stage in zygotes lacking maternal TH2A/TH2B. H3K27me3 is a mark of repressive chromatin, and its formation is catalyzed by the enzymatic subunit Ezh2 in the Polycomb repressive complex 2 (PRC2) (Margueron and Reinberg, 2011). In WT zygotes, the H3K27me3 signals on the paternal genome started to increase at syngamy, after the PN5 stage, indicating that a loss of maternal TH2A/TH2B accelerates the formation of repressive chromatin in the paternal genome. In zygotes lacking maternal TH2A/TH2B, the level of H3K14ac on the maternal genome at PN0 was much lower than that of the control WT zygotes, but this decrease was recovered at PN2.

Note that when the authors add TH2A, TH2B, and Npm to their reprogramming cocktail they refer to those 3 factors together as “BAN”. A histone chaperone, nucleoplasmin (Npm), seems important as well. A phosphorylation-mimic form of Npm (P-Npm)  enhanced reprogramming involving BAN.

It’s interesting that BAN plus Klf4 and Oct4 (referred to as KOBAN…made me think of Kurt Cobain) was sufficient for reprogramming.

With global chromatin opening factors like these histone variants and Myc, one of the tricky parts is looking at gene expression since it is possible for these factors to indirectly boost expression of many genes via their widespread chromatin activating function. However, the authors report some specific gene expression effects in Figure 5. They saw relatively discrete changes with BAN added to the mix with just a few hundred specific genes impacted. 22% overlapped with so-called “initiation” genes found previously to be turned on during early reprogramming.

They also used ChIP-Seq to identify the genomic location of TH2A and TH2B and found that its binding specifically to the X chromosome was greatly enriched during reprogramming, but what it is doing there and how that mechanism then feeds into reprogramming remain interesting open questions.

You can see their model above at the top of the post. One of their more notable conclusions is that the histone variants make cellular reprogramming more similar to SCNT:

the mechanism of reprogramming using TH2A/TH2B in conjunction with OKSM more closely resembles SCNT than conventional reprogramming and that TH2A/TH2B also play a functional role in reprogramming by SCNT.

TH2A and TH2B also seem to play roles in the global chromatin activity states of the maternal and paternal genomes. The authors further point out that another booster of reprogramming a couple years back, Glis1, is  also enriched in expression in oocytes and zygotes.

Overall I thought this paper was very strong and opens up a lot of interesting mechanistic questions related to chromatin during reprogramming.