Wednesday, July 23, 2014

Why is this IL-33 paper so "alarmin"-gly bad?

Some papers just make no sense, especially when they appear in top journals, like Nature. What's going in editor's mind when they are green-lighting this type of research for publication?

Here is a new example for such paper. It was recently published in Nature (1). This study comes from Fiona Powrie's lab in UK. She became well-known for her studies of experimental colitis model in rats or mice upon adoptive transfer of naïve T cells (then, in 90's, called CD45RBhi). Her research is mainly focused on interplay between naïve T cells and Foxp3+ Tregs in colitis models.

This study is another variation of this approach.

First, the authors made an observation that gut associated Tregs express IL-33 receptor, while few Tregs from spleen or lymph nodes do.

Next, in vitro test showed that combination of IL-33 and TGFbeta-1 augments de novo Foxp3+ Treg differentiation (IL-33 alone had no effect).

To assess in vivo role of IL-33 signaling in Tregs, the authors first generated mixed bone marrow chimera mice using wild-type and IL-33 receptor-deficient marrow cells (also called ST2 -/-). Next, colitis was induced by combination of treatment with Helicobacter hepaticus and anti-IL-10R injection. Analysis of number of colonic Tregs derived from IL-33 receptor-deficient marrow cells (which can not respond to IL-33) were reduced 2-fold compared to wild-type Tregs. In addition, IL-33 receptor-deficient Tregs expressed reduced amount of Foxp3 compared to wild-type Tregs (1000 vs. 800 MFI).

Another set of in vivo experiments showed that IL-33 receptor-deficient Tregs were less potent in preventing colitis induction by naïve T cells. At 8 weeks post transfer, IL-33 receptor-deficient Tregs lost much of Foxp3 expression. Of note, in vitro IL-33 receptor-deficient Tregs were as potent as wild-type Tregs.

To connect this new observation with their previous studies about IL-23 and colitis, the authors showed that IL-23 blocks signaling by IL-33 in Tregs.

Finally, another set of in vivo adoptive transfer experiments into RAG1/IL-23R double knockout mice showed that in even in absence of IL-23 signaling in the host, IL-33 receptor-deficient Tregs were less potent in preventing colitis induction. 

The data in Fig. 4d, however, are difficult to interpret. If IL-23 was blocking Tregs induction solely through IL-33 inhibition, absence of IL-23 should have had two fold outcome:

(a) induced much potent wild-type Tregs and

(b) IL-33 receptor-deficient Tregs should become more potent in preventing colitis

However, neither outcome has been observed, that is, in my opinion, very unusual and unexplainable result and questions the relevance or importance of all other data.



  1. Although that would have been nice to include a control set of mice (transfer of WT or ST2 KO in Rag ko), I don't see where the problem is...
    In the absence of IL-23, IL-33 can do its job, which is to induce Tregs (white bars), but when IL-33R Ko t cells are transferred, it can no longer do it, so there is less Tregs and a worse clinical score (black bars).
    I don't understand why you say that IL-33R KO Tregs should be more potent to prevent colitis...

  2. Thanks for comments.

    I agree.

    RAG KO recipient would have been an important control in Fig.4, especially since they have used such recipient in Fig.3 experiment.

    Actually, in my opinion, this omission is already problematic for Nature level paper.

    What I meant, however, was that absence of both IL-23 and IL-33 signaling should cancel out each other. Why would Treg induction require IL-33 in absence of IL-23? Just more Tregs?

    Since colitis score is very similar in Fig 3 and Fig 4, even if we take into account that they are very different type of experiments, I thought this paper required more scrutiny by reviewers.

  3. Although it is difficult to make a straight comparison because of the different genetic background of used recipient mice, it looks to me that the wild-type Tregs in IL-23/Rag1 double KO mice have definitely done a better job (colitis score fig. 4e vs. fig. 3a). I think though that there are more little problems with this paper (e.g. try to match some of the qPCR results (fig. 4b) with Western Blot data (fig. 4g & c)). And if you go trough the supplementary data, all the overstatements mentioned in the main text become really evident…