New Nature papers debunk STAP cells

Today marks nearing the completion of a full circle for one of science’s biggest controversies: the STAP cell fiasco. Today STAP cells are completely refuted with the publication of two new papers in Nature and we know much more–with some notable gaps still–about what went wrong.

In January of last year, an international team of collaborators from RIKEN in Japan and Harvard/Brigham & Women’s Hospital (including the lab of Charles Vacanti where the STAP idea reportedly originated) here in the US published two Nature papers making the extraordinary claim that ordinary cells could be reprogrammed into embryonic stem cell (ESC)-like cells.

And it could be done simply, cheaply, and quickly using various forms of cellular stress including low pH. I was highly skeptical when I read the papers, but tried to keep an open mind. This sounded cool, even if also too good to be true.

I published a review of the papers here on this blog on the day they were published and I included six key open questions that would be required to assess the real impact of these papers. Over the next few weeks I posted an increasingly skeptical series of posts questioning STAP.

Others in the larger community including anonymous scientific sleuth JuuichiJigen and some on PubPeer were skeptical as well. In fact, they started noticing issues with the data and text of the papers.

RIKEN and Nature began investigations. Ultimately the papers were retracted in relatively quick fashion. While a lot of harm was done even so and tragedy would strike later, the rapid refutation of STAP attenuated the overall damage.

For more background on the key STAP events check out this comprehensive STAP history timeline. Ken Lee’s lab took the lead in scientific refutation of STAP and published their work in F1000 here after Nature rejected it under unclear circumstances.

I also started a novel, but admittedly somewhat basic attempt at crowdsourcing global efforts at STAP replication. Very quickly we came to a consensus that autofluorescence was likely a key stumbling point for the STAP papers as the authors probably misinterpreted it as real signal from a GFP pluripotency reporter.

Suspicions grew elsewhere that STAP cells might really be ESCs or some other pluripotent stem cells, possibly mixed with trophoblastic stem cells (TSC). Ultimately, STAP first author Haruko Obokata was found by RIKEN to have committed misconduct and she is no longer working at the institution. RIKEN underwent a big shakeup as a result of STAP as well. STAP co-author and highly respected biologist Yoshiki Sasai committed suicide, which was one of the most tragic and sobering events I’ve seen in science during my career. In Japan there had been a media frenzy on the STAP problems. In the US things on the STAP front were and continue to be quieter. As recently as about a year ago, Vacanti and co-author Koji Kojima publicly expressed complete confidence in STAP and put up a refined protocol on the web.

So what was the real deal with STAP?

Today Nature published two articles thoroughly refuting STAP cells and providing some further insights.

In one of the papers, STAP cells are derived from ES cells, the authors used whole genome sequencing (WGS) to examine archived STAP cell-related samples and other cells present in the laboratories where the STAP work was conducted. Using essentially a form of genomic fingerprinting, the team reports conclusive evidence that STAP cells were in actuality ESCs:

In summary, our investigations based on WGS of STAP-cell related materials reveal that all of these materials are derived from previously established ES cell lines and refute the evidence shown in the two Nature papers that cellular stress can reprogram differentiated cells into pluripotent cells.

You can see Figure 1b from this study showing the WGS comparison that the genomic characteristics of various cell lines.

STAP refutationThe matching patterns between two STAP-derived lines FLS3 and CTS1 and the supposedly unrelated FES1 ESC line are particularly striking. It now seems almost certain that a number of STAP cells are in reality FES1-related ESC lines and that the STAP cells were not created by cellular stress.

The other new paper from another team, Failure to replicate the STAP cell phenomenon, comes to similar conclusions and further clarity arises:

“In summary, our replication attempts and genetic analysis indicate that existing STAP protocols are neither robust nor reproducible. To substantiate future claims of reprogramming and alternative states of potency, we urge a rigorous application of several independent means for validating functional pluripotency and genomic profiling to confirm cell line provenance. Ultimately, the essential standard of robustness and reproducibility must be met for new claims to exert a positive and lasting influence on the research community.”

This second team led by George Daley at Brigham and Women’s spans the globe, but importantly they did some of the work actually in Vacanti’s lab, still finding no evidence that STAP is real. They wrote, “Working within the Vacanti laboratory where the concept of STAP cells originated, and assisted by a co-author of the STAP papers…”

Seven laboratories were involved in this second STAP replication effort: Daley, Deng, Hanna, Hochedlinger, Jaenisch, Pei and Wernig. This is an all-star team of stem cell research labs.

One bottom line from the paper is that this team collectively worked very hard to try to get STAP to work, but it didn’t:

“In summary, 133 replicate attempts failed to document generation of ES-cell-like cells, corroborating and extending a recent report.”

Like the other team, these scientists analyzed the STAP cells including their genomes. They found inconsistencies between their new findings and the claims in the original STAP papers:

“In the original STAP reports, the authors stated that they mixed CD451 cells from male and female mice owing to the small number of CD451 cells retrieved from individual neonatal spleens. However, our analysis indicates that CD451 cells were female, whereas the derived cells (STAP cells, STAP stem cells and FI-SCs) were all male, a clear inconsistency.”

These authors also found indications of trophoblastic stem cells (TSC) being mixed into the STAP samples. TSC may explain the reported totipotency of some derivations of supposed STAP cells.

Nature itself explained why it published these new papers (in the Brief Communications Arising or BCA format):

“Why is Nature publishing these pieces? The main reason is to update the scientific record. The wording of the STAP retraction notices left open the possibility that the phenomenon was genuine. It said: “Multiple errors impair the credibility of the study as a whole and we are unable to say without doubt whether the STAP-SC phenomenon is real.” The two BCAs clearly establish that it is not.”

We are just about, but not quite at the end of the STAP story it seems. In my opinion there is still more to be learned about what went so wrong. How did the ESCs and in some cases TSCs end up in the cell culture mix? Accidental contamination? Intentional attempt to bolster the seductive hypothesis?

We may never know, but today there is a great deal more clarity overall at least.

The publication of these two new papers is a very positive step, but it is important to stress that absent post-publication review, rapid and open team science, and social media efforts, the STAP cell myth may have continued to have been believed by many in the research world until this day when these debunking papers were published. That delay would likely have caused immeasurable damage. Thus, there were important roles both for traditional scientific correction via journals and new, transformative types of rapid post-publication review.

TGIF: Stem Cell Stream of Consciousness for the Week

It seems like Fridays are always a good time for reflection.

It’s not like the scientist’s week ends on Friday sad to say, but still at least we can pause momentarily and reflect on the maelstrom that is our lives, while cracking open another diet Coke or pouring another cup of coffee.

This week was a great one for getting into some intense discussions with other scientists both in person and electronically.

CRISPR the talk of the biology town. Everyone is talking about CRISPR-Cas9. There are many subtopics. I’m just going to throw them on the table in a stream of consciousness here. Gee, CRISPR is not as specific as I thought. Uh oh, CRISPR is harder to get to work than I thought (as portrayed in early papers). CRISPR is fricking awesome. Oh, **** , people are going to CRISPR human embryos? The people who innovated CRISPR-Cas9 like Jennifer Doudna are amazing and deserve a Nobel Prize.

CIRM change leaves some feeling unsettled. There is starting to be a feeling of disconnect, especially amongst more basic stem cell researchers, with CIRM.

CIRM just isn’t–at least so far in the existence of CIRM 2.0–funding basic or early translational stem cell research much or at all any more. Maybe it will in the future as there have been hints about that, but support from CIRM in a major way for important, basic stem cell research is probably basically over. The wildly successful and positive CIRM Bridges Program is nearing an end. The CIRM T32 training program are winding down. Many of the early-mid-stage CIRM grants are ending.Lujan paper

There’s definitely a feeling in the California stem cell community of CIRM kind of having moved on from a whole class of stem cell researchers barring those fortunate few who still have grants. Maybe there can be a future reunion.

CDI Buyout Big Bucks. Fujifilm shelled out almost a 1/3 of a billion dollars for Jamie Thomson’s Cellular Dynamics International (CDI). I guess IPSC are really worth big bucks, although CDI isn’t so IPSC-specific anymore. Regenerative Medicine is now proving its worth big bucks. Billions and billions?

Cool New Wernig Lab paper in Nature, but what does it mean? I saw that new Wernig lab Nature paper with Garry Nolan’s uber-fancy FACS-mass spec data. Those SPADE analysis images that look like trees branches into rainbowy fractals sure are pretty. Is it like some kind of heatmap-esque, principle component, multidimensional analysis coolness? Yeah, I’m going to have to read this Wernig lab paper more carefully.

Over at the Stanford media the wonderful Krista Conger boiled the paper down to say that there is a defined “transition state” during reprogramming.

Transdifferentiation makes a major advance: direct reprogramming of fibroblasts to oligodendrocyte progenitors

Back-to-back papers (here and here) in Nature Biotechnology report the transdifferentiation (now often simply referred to as “direct reprogramming”) of plain old fibroblasts into brain cells called oligodendrocyte progenitor cells (OPCs).

OPCs are a remarkably useful kind of brain cell that generates myelin, which insulates nerves. OPCs are thought to have great therapeutic potential for a number of pathological conditions including Multiple Sclerosis (MS) and spinal cord injury (SCI).

The two teams included one from Case Western Reserve University School of Medicine led by Paul J Tesar (Najm, et al paper) and one from Stanford (Yang, et al paper) led by transdifferentiation guru, Marius Wernig. The teams worked independently but made very similar, complimentary discoveries.

transdifferentiation OPCs

Both teams honed in on transcription factors that are specifically expressed in OPCs and the oligodendrocyte lineage or when deleted cause defects in this lineage.

Najm, et al screened using the following factors: Olig1Olig2Nkx2.2Nkx6.2Sox10ST18Gm98 (Myrf) and Myt1.  They found that all 8 factors when introduced together could make OPCs, but that Sox10, Olig2, and Nkx6.2 could do the same thing by themselves without the other five factors.

Yang, et al found that Sox10, Olig2, and Zfp536 could directly reprogram fibroblasts into OPCs after screening through the following: Ascl1, Gm98, Myt1, Nkx2.2, Nkx6.1, Nkx6.2, Olig1, Olig2, Sox10 and Zfp536.

Taken together these findings would suggest that Sox10 and Olig2 are the most important OPC-inducing factors. It is interesting that the teams each found a distinct 3rd factor makes up an effective 3-part cocktail.

It is particularly notable that Najm’s cocktail uses the homeodomain transcription factor Nkx6.2 because it seems that the Yang team tried Nkx6.2 as well, but in combination with Sox10 and Olig2, it didn’t work in their hands. Instead, a zinc-finger transcription factor Zfp536 did the trick for Wernig’s team.

The difference may be explainable by other different conditions and assays used in the two labs.

Part of Figure 2 from the Najm paper (above) shows that in the presence of Dox, which turns on the 8 transgenes, the morphology changes from normal fibroblasts (spindly in a) to the characteristic bipolar morphology of OPCs (b). With differentiation, the cells adopt the beautifully intricate morphology of oligodendrocytes, stain for the GFP (green) marker indicating the transgenes are “on”, and also importantly stain in red for Maltose Binding Protein (MBP), a marker of mature oligodendrocytes.

Very cool!

In both cases with the two teams, the studies were done using rodent cells and each of papers notes in the discussion that human studies are critical. Wernig’s team stated:

Given the strong clinical interest in OPCs for regenerative therapies, one of the most important next steps is to translate our findings to human fibroblasts. Based on our experience with iPS and iN cells, we predict that generation of human iOPCs is possible but may require additional reprogramming factors, such as other transcriptional regulators or microRNAs

I see these papers as a major advance in transdifferentiation technology. In theory if these technologies can be adapted for human use, then transdifferentiation can be used to make OPCs from any given patient that could be given back to that same patient as an autologous transplant. Thus, these studies have huge clinical implications.

I am happy to say that this discovery also now makes me have a perfect 100% record so far for my 2013 stem cell predictions (you can check them out here).

I am now correct on #8, #6, #3 (so far), and #1.

Fingers crossed on the rest.