Evidence favoring the biophysical model for Hox gene collinearity

On March 29, 2011 Patrick Tschopp wrote:

Dear Spyros,

thank you for your message. I have read already your recent review, in fact Denis gave me the print-out you sent him. However, I also stand by the conclusions and the model we deduced from our series of experiments.

The Evx3-Hoxd13 region has been investigated in quite some details, with the recent inversion as well as deletions encompassing Hoxd13. If only this stretch of DNA was deleted, no premature expression was observed, only when fragments more centromeric to Evx2 were affected (see Kondo and Duboule 1999; Tschopp and Duboule 2011).

A clean deletion of this intergenic region would require the generation of two new loxP sites, thus tedious ES cell work and at least 2 -3 years of follow-up. As I will be leaving the lab in less than two months, I will not be able to consider either one of the experiments you suggested!

Thanks anyway for your continued interest in our work.

Sincerely,

Patrick

------

The above email gave me the opportunity to read carefully the Kondo and Duboule paper (Cell 97, 407-417 (1999)) and this was a gratifying revelation.

In this paper, among other experiments, K and D describe the following genetic deletions in the Hoxd cluster:

Deletion del0: the posterior region of the cluster is deleted 

                                          ↑---Hoxd13------ Hoxd12------- Hoxd11---↑

Deletion delII: together with del0, the exterior region --Evx2----------- is deleted:

                       ↑--Evx2----------------Hoxd13------ Hoxd12------- Hoxd11---↑

K and D examine the expressions of Hoxd4 and Hoxd10 after deletions del0 and delII for transgenic mice embryos at stage E7.5 to E8.5. Their results are shown in Fig.6 of their 1999 Cell paper and they conclude (p. 412) “ …that both  Hoxd4 and Hoxd10 were activated prematurely in the delII  configuration. Unexpectedly again, analysis of Hoxd4 and Hoxd10 expression in older del0 and delII fetuses did not show any deviation from their normal profiles (not shown), suggesting that the early deregulation was not maintained subsequently”.

Let us make an integrated analysis of the above unexpected results in the framework of the biophysical model:

1. The wild type expression of Hoxd4 at stage E8 is observed posteriorily in the embryo while this expression is missing (at the same stage) in mutant embryos with a posterior Hoxd deletion (see Fig. 6D in P.Tschopp et al. (2009) PloS Genet. 5  (3) ).

This observation was explained recently (S. Papageorgiou (2011) Develop. Growth Differ. 53, 1-8). [Let me repeat in short the explanation: in a posterior deletion, the total negative charge N of the cluster is reduced. In order to compensate this reduction, the positive charge P of the surroundings must increase and this causes a retarded posteriorization of the expression. Hence, at the stage E8 the mutant expression of Hoxd4 has not appeared yet]. At E7.5 to E8.5 and for del0, the Hoxd4 expression is absent –in contrast to the wt Hoxd4 expression.

2. A basic hypothesis of the biophysical model is the representation of the Hox cluster by an elastic spring whose fixed end is located inside the chromosome territory and its free end can be pulled toward the interchromosome domain. The fixed end lies between the 5’end of the cluster and Evx2. When the fixed end of the spring is cut-off, the spring becomes loose and can be pulled and expanded with smaller than normal forces –which means prematurely.

In  delII, besides del0, an extra region is deleted (---Evx2-------) which lies outside the Hoxd cluster. This is the region of the fixed end of the spring. Therefore in delII as compared to del0 a premature gene expression is expected. This is confirmed and depicted in Fig.6 of the K and D (1999) paper. This ‘unexpected’ premature activation is naturally derived from the biophysical model. To my knowledge, only the biophysical model predicts this behavior of the Hoxd transgenic expressions.

3. In the experimental proposals (see my blog entry March11, 2011), the aim of Experiment A is to test whether the Hoxd expressions are prematurely expressed after the deletion of the   --Evx2---------    region. It is ironic that this model expectation of premature activation was already confirmed in retrospect by Kondo and Duboule in 1999! Therefore, Experiment A is not necessary anymore. The proposition for Experiment B still holds.

Spyros Papageorgiou

Email:spapage@bio.demokritos.gr

A sad sequel to the ‘correspondence’ with D. Duboule

D. Duboule wrote back in the same angry style:

Spyros,

Let me add (for your blog) that making public a private mail from me re-enforces my views about your complete lack of education.

D

Prof. Denis Duboule

Dept of Genetics and Evolution

Faculty of Sciences, University of Geneva

The above message raises an issue that deserves an answer:

1. In scientific matters, there is no room for privacy or secrecy.

2.  In the case of an offence, the right of privacy can be asked (optionally) by the defender only. Otherwise, the claim of privacy by the assailant indicates that he has not the stamina to defend his position in public.

Spyros Papageorgiou 

A disturbing 'correspondence' with Dr D. Duboule

When my paper (Develop.  Growth & Differ. (2011) 53, 1-8) was accepted for publication, I sent a preprint to Prof. Denis Duboule with the following covering letter.

Prof. D. Duboule

University of Geneva

Geneva, Switzerland

                                                                         5 November 2010

Dear Dr. Duboule,

      I enclose a preprint of my work that is going to be published in Development, Growth & Differentiation.

      My work is mainly based on the experiments of your group. However, the interpretation of your results that I propose in the biophysical model differs from your two-phases model.

        Any comments are welcome.

Sincerely,

S. Papageorgiou

Institute of Biology, NCSR ‘Demokritos’, Athens, Greece

Email:spapage@bio.demokritos.gr  

As expected, I received no answer from DD.

Last month, an article by P.Tschopp and D. Duboule appeared in Dev. Biol. (2011), 351, 288-296 and, as a result, I thought of two experiments that could test both models (see my blog Two experimental proposals…). Since DD does not bother answering my letters, I sent my proposals to Patrick Tschopp (a co-author and PhD student of Duboule):

                                                                            

                                                                                                  28 March, 2011

Dear Patrick,

I know your supervisor has strong objections against my biophysical model for Hox gene collinearity. Probably this is understandable since he believes he has found the ‘true’ explanation of collinearity.

In the attached propositions however, two experiments are described which can distinguish which model is correct (if any at all). In your lab these experiments are easily done.

In the quest for scientific truth, it is worth performing the above experiments even if you are convinced that the results will only confirm your predictions.

Looking forward to hearing from you.

Best wishes,

Spyros

S. Papageorgiou

P.Tschopp answered next day politely explaining why he could not undertake this task:

Dear Spyros,

thank you for your message. I have read already your recent review, in fact Denis gave me the print-out you sent him. However, I also stand by the conclusions and the model we deduced from our series of experiments.

The Evx3-Hoxd13 region has been investigated in quite some details, with the recent inversion as well as deletions encompassing Hoxd13. If only this stretch of DNA was deleted, no premature expression was observed, only when fragments more centromeric to Evx2 were affected (see Kondo and Duboule 1999; Tschopp and Duboule 2011).

A clean deletion of this intergenic region would require the generation of two new loxP sites, thus tedious ES cell work and at least 2 -3 years of follow-up. As I will be leaving the lab in less than two months, I will not be able to consider either one of the experiments you suggested!

Thanks anyway for your continued interest in our work.

Sincerely,

Patrick

-----

The same day I received an e-mail from D. Duboule himself in relation to my message to P. Tschopp:

Spyros,

I find your mail insulting- Why dont' you mind your own business?

If you have good ideas, go for it and leave us alone-

You just lack a sense of education-

D

Prof. Denis Duboule

Dept of Genetics and Evolution

Faculty of Sciences, University of Geneva

Sciences III, 30, Quai Ernest-Ansermet

1211 Genève 4, Switzerland

Tel  (41) 22 3796771; Fax (41) 22 3796795 

http//zoologie.unige.ch

http//www.frontiers-in-genetics.org

Administration: Corinne Matthey-Ebener 

Tel. (41) 22 7026770 (6785)

Corinne.Matthey-Ebener@unige.ch

-----

I think this is too much!

I answer publicly to DD since my arguments might be of some interest to other members of the scientific community.

1. The readers can judge for themselves whose writings are insulting and rude.

2.  Performing ground breaking experiments is one thing, their proper explanation is

     another. In other disciplines (e.g. Physics) Experiment and Theory (although in

     coherent cooperation) have been separated a long time ago. Probably E. Fermi was

     the last great Physicist- experimentalist and theoretician at the same time. Many

     excellent biologists should accept this natural metamorphosis of their field, since

     the quantitative analysis of their complex data requires the involvement of other

     branches of Science like Mathematics and Physics. 

     Hand waving qualitative ‘models’ give place to accurate estimates and elaborate

     calculations. Experiment, however, remains the indispensable foundation of the

     scientific edifice. This sine qua non prerequisite underlies all efforts along the path

     from hypothesis to Theory. Any criticism of proposed ideas, as long as it is based

     on scientific arguments alone, is welcome, constructive and free to all.

3.  Coming to the problem of Hox gene collinearity, it is more than 10 years that I try

     to understand this enigma. I have formulated a model, the biophysical model,         

    which explains satisfactorily - to my taste- the existing data. The last years genetic

    engineering experiments were performed almost exclusively in Duboule’s lab.

    Duboule and co-workers have proposed the ‘two-phases model’ which, I think,

    does not convincingly explain the data. It is ironical that all their new results fit

     better in the biophysical model whereas many of their findings remain unexplained

     by their model. I still believe it is worth performing the experiments I propose

     above or arrange other similar setups to explore the behaviour of the posterior region

     of the Hoxd cluster and compare it to the fixed end role of an elastic spring.

     Passing over D’s out of temper response, I personally feel thankful to the work

     of his team for their experimental achievements since their findings helped me

     substantially in the formulation of my model.  

Spyros Papageorgiou    

Proposal for experiments to test the mechanism of HOX gene collinearity

Two experimental proposals clarifying the mechanism of HOX gene collinearity

    Hox gene collinearity is a surprising phenomenon observed during axonal development of embryos. This phenomenon is not satisfactorily understood as yet. In the last years, a series of genetic engineering experiments on mice embryos were performed in Duboule’s laboratory (ref. 1-3). These interesting experiments reveal some unexpected facets of Hox gene collinearity. In order to describe their findings, Duboule and coworkers proposed a ‘two-phases model’ (T-P M). According to this model (ref. 1-3), in the early phase (up to about stage E9.5) two regulatory mechanisms come into play, one positive acting from the anterior side (3’) of the Hox cluster and a repressive one acting from the centromeric side of the cluster. In their latest paper (ref. 3) the authors claim that the repressive regulation is related to a ‘landscape effect’ from the extended region centromeric to the cluster.

       I have proposed a quite different model, the biophysical model (BM), which explains Hox gene collinearity satisfactorily (ref. 4-6).  The recent genetic engineering experiments are also well reproduced by the biophysical model which is based on the hypothesis that physical forces decondense the Hox cluster and pull the Hox genes sequentially from inside the chromosome territory toward the interchromosome domain. Thus the Hox cluster behaves like an elastic spring whose posterior end is fixed inside the chromosome territory and the anterior end (3’) is free and can be pulled toward the interchromosome domain (ref. 4-6). In this mechanical representation, the fixed end of the spring is essential. It is natural to assume that it lies in the small region (Ev-13) between Evx2 and the last gene Hoxd13 of the cluster.     

     In what follows two simple experiments are proposed in order to distinguish which of the two models (T-P M and BM), if any, is correct. For each of the proposed experiments the above models predict divergent results. It will be therefore interesting to compare the model expectations to the experimental findings.

1. Experiment A:

With the methods described in ref. 1-3, the small posterior region (Ev-13) between Evx2 and Hoxd13 is deleted.

a) According to  T-P M, this small deletion should not affect substantially the spatial and temporal expressions of Hoxd13, Hoxd12, Hoxd11, …No difference between wild-type and mutant expressions is expected.

b) In contrast, according to the BM, this small posterior deletion destroys the spring character of the Hox cluster since its fixed end is removed and the spring remains loose and can be easily shifted by small forces as explained in ref. 5-6. Therefore, Hoxd13, Hoxd12, Hoxd11, … should be prematurely over-expressed in the early phase.     

2. Experiment B:  Following the method described in ref. 3 an inversion is performed centromeric to Evx2 (not including the intergenic region between Evx2 and Hoxd13).

a) The Hoxd13, Hoxd12, Hoxd11… expressions in the early phase should be comparable to the observed expressions of the experiment presented in Fig. 3 of ref. 3 since, in both experiments, the inverted centromeric landscape is almost the same. According to T-P M, this inverted centromeric region is responsible for the observed abnormal expressions.  Therefore, we expect the mutant expressions to differ significantly from the wild-type embryo expressions.    

b) The biophysical model predicts at the early phase a quite different behavior: the elastic spring with its fixed end remains intact, therefore its function is normal and this centromeric inversion does not affect the mutant Hoxd gene expressions. As a result, the mutant Hoxd13, Hoxd12, Hoxd11… expressions should not differ from the wild-type expressions.

 I think the above experiments constitute crucial tests for both T-P M and BM.

References  

1. B. Tarchini and D. Duboule (2006). Developmental Cell 10, 93-103.

2. P. Tschopp, B. Tarchini, F Spitz, J. Zakany and D. Duboule (2009). PloS Genet. 5 

    (3).

3. P. Tschopp and D. Duboule (2011). Dev. Biol. 351, 288-296.

4. S. Papageorgiou (2006). Int. J. Dev. Biol. 50, 301-308.

5. S. Papageorgiou (2009). Hum. Genomics 3, 275-280.

6. S. Papageorgiou (2011). Develop. Growth Differ. 53, 1-8.

-----

Spyros Papageorgiou

Institute of Biology, NCSR ‘Demokritos’, Athens, Greece

Email: spapage@bio.demokritos.gr