Guest post by Paul Knoepfler, cross-posted on ipscell.com
What is “radically deterministic” cellular reprogramming?
Jacob Hanna’s lab recently published a very high profile Nature paper on nearly 100% efficient iPS cell formation based on targeting a factor called Mbd3. This super efficiency is a major departure from past reported efficiencies.
This morning here at the Till & McCulloch Meetings in Banff, Hanna gave an intriguing talk telling us more about his lab’s efforts in this area. He also fielded some incisive questions from the audience (including 2 from me) after his talk.
A dominant model in the reprogramming field recently has been that there are two main types of reprogramming elements or phases: stochastic (random) and deterministic (in a stepwise, orderly fashion).
In his talk today Hanna postulated that there are key drivers and barriers to the process and further that by manipulating them perhaps we could make reprogramming more deterministic. He analogized the drivers to accelerators in a car, while the barrier factors were the brakes.
In terms of data, he first talked about a reprogramming driver called Utx1, a Jumanji family histone demethylase. Utx1 promotes reprogramming by removing K27me3 marks from pluripotency genes (see a Nature paper from the Hanna lab on this work here).
How does this work?
Hanna indicated that Utx1 directly binds to the key reprogramming factors Oct4, Sox2, and Klf4 (OSK) and recruits them to derepress silenced pluripotency genes. Presumably in this kind of model, therefore, OSK overexpression would have to occur prior to Utx1’s driver function being invoked because somatic cells don’t expression OSK.
Hanna shifted gears and focused in the remainder of his talk on the factor Mbd3 as a suppressor of cellular reprogramming.
Most people making iPS cells accept the notion that reprogramming has been in the past a mixture of stochastic and deterministic phases. However, the Hanna Lab paper (Rais, et al.) found that by lowering the levels of or eliminating Mbd3 they could shift things to be remarkably deterministic.
I reviewed that paper in a previous blog post and noted some open questions about the mechanisms involved in exactly how loss of the Mbd3 protein leads to such a high level of reprogramming efficiency.
Hanna went through the work in his paper on Mbd3 in his talk. Interestingly, as with Utx1, Hanna argued for a model in which Mbd3 also directly interacts with OSK too and he specifically mentioned that it does this via its Mbd protein domain.
I asked Hanna during the Q&A after his talk whether Mbd3 and Utx1, since they both bind OSK, could be in a complex together as well.
He said that issue remains unclear, but it is something they have thought about too. I also asked him if endogenous Mbd3 and endogenous Utx1 bind endogenous OSK. He said they do form endogenous complexes.
He also mentioned in his reply that he saw my critique of his Mbd3 paper on my blog where I had mentioned my concern that perhaps Mbd3 only binds OSK when they are overexpressed to supraphysiological levels and he said endogenous interaction was shown in supplemental data.
So I took a look at the paper after his talk to refresh my memory. I had recalled in my memory at least that the supplemental data showed another NURD complex member (Chd4), but not Mbd3 binding to the reprogramming factors (and importantly binding to not endogenous but rather exogenous reprogramming factors induced by Dox). Indeed that is what can be found in Supplemental Figure 9 here of his paper. Further, one should note that the levels of reprogramming factor binding to Chd4 are rather low overall. As a result, I’m not a full believer in this proposed mechanism.
Hanna called reprogramming without Mbd3 by the moniker “radically deterministic” reprogramming, which I thought was notable as a new catch phrase.
From the audience came some tough, but very smart questions after mine.
Someone asked if loss of Mbd3 would predispose to cancerous cellular behavior and the answer seemed to be “yes” so that’s a potential concern, however all reprogramming factors seem to be linked to cancer one way or another. Thus, as Hanna added, the key may be the use of transient molecular interventions such as via small molecules that do not leave genetic fingerprints in the resulting iPS cells. This point resonated with Hongkui Deng’s talk a bit later today on all-chemical reprogramming.
Another audience member who I believe was the PI of a lab pointed out that the Dox-inducible secondary reprogramming system that Hanna used was, in his lab’s hands, already 30-80% efficient without doing anything to Mbd3. I don’t think this point was resolved. It is also notable that in a separate talk today, Andras Nagy also highlighted that reprogramming using the so-called “secondary” system of fibroblasts made from mice made from iPS cells (again with no Mbd3 intervention) had extremely high efficiency.
Finally, another person asked if Hanna had tested whether his reprogrammed cells with Mbd3 loss-of-function were “truly pluripotent” as in, as she put it, “able to form murine chimeras”. Hanna was confident that the Oct and Nanog reporters faithfully showed reprogramming to true pluripotency, but I wasn’t clear if they had done rigorous chimeric mouse assays to support the claim that 100% of cells are reprogrammed to true pluripotency.
Overall, despite the tough questions and some caveats noted above, I thought it was an outstanding talk and this is clearly very important work.
Yes, frankly, I’m not yet convinced that Mbd3 and Utx1 both function simply by directly and individually binding at the protein level with OSK, but I’m sure the mechanisms will be ironed out. I’ll be very curious to follow the work from Hanna’s lab (and possibly others) as they delve more into the molecular mechanisms. Much of this is going to depend, as he indicated, on the nature of these powerful protein complexes.