Development

We have applied the logical modelling framework to the regulatory network controlling T-lymphocyte specification. This process involves cross-regulations between specific T-cell regulatory factors with factors driving alternative differentiation pathways, which remain accessible during the early steps of thymocyte development. Many transcription factors needed for T-cell specification are required in other hematopoietic differentiation pathways, and are combined in a fine-tuned, time-dependent fashion to achieve T-cell commitment.
Using the software GINsim, we integrated current knowledge into a dynamical model, which recapitulates the main developmental steps from early progenitors entering the thymus up to T-cell commitment, as well as the impact of various documented environmental and genetic perturbations. Our model analysis further enabled the identification of several knowledge gaps.

The associated notebook can be loaded using the CoLoMoTo notebook docker image (see http://www.colomoto.org/notebook).

Jupyter Notebook: Tdev_notebook_2nov2019.ipynb

This Boolean model displays the subtle role of miR-9 in the course of neural specification in zebrafish.

Figure 1: Green arrows represent positive regulations, whereas red T arrows represent inhibitory interactions. Dotted lines represent suspected (but not yet molecularly characterized) direct or indirect interactions, which prevent the expression of a gene in the progenitor or neuronal precursor states, such as elavl3/HuC inhibition in progenitors.

The node P denotes a proliferating progenitor state (P=1, N=0). It is defined by the expression of Her6 and/or Zic5. The node N denotes the commitment of a progenitor into a neural precursor (P=0, N=1).
By inhibiting genes with opposite effect on neural differentiation, miR-9 activity generates an ambivalent state (P=0, N=0) poised for responding to both progenitor maintenance and commitment cues.

Model simulations qualitatively recapitulate all the experimental results presented in Coolen et al (submitted), for the wild-type as well as for various mutant situations (including loss-of-functions, ectopic gene expressions, and miR-9 target protection by morpholinos).

Background
Primary sex determination in placental mammals is a very well studied developmental process. Here, we aim to investigate the currently established scenario and to assess its adequacy to fully recover the observed phenotypes, in the wild type and perturbed situations. Computational modelling allows clarifying network dynamics, elucidating crucial temporal constrains as well as interplay between core regulatory modules.
Results
Relying on a comprehensive revision of the literature, we define a logical model that integrates the current knowledge of the regulatory network controlling this developmental process. Our analysis indicates the necessity for some genes to operate at distinct functional thresholds and for specific developmental conditions to ensure the reproducibility of the sexual pathways followed by bi-potential gonads developing into either testes or ovaries. Our model thus allows studying the dynamics of wild type and mutant XX and XY gonads. Furthermore, the model analysis reveals that the gonad sexual fate results from the operation of two sub-networks associated respectively with an initiation and a maintenance phases. At the core of the process is the resolution of two connected feedback loops: the mutual inhibition of Sox9 and ß-catenin at the initiation phase, which in turn affects the mutual inhibition between Dmrt1 and Foxl2, at the maintenance phase. Three developmental signals related to the temporal activity of those sub-networks are required: a signal that determines Sry activation, marking the beginning of the initiation phase, and two further signals that define the transition from the initiation to the maintenance phases, by inhibiting the Wnt4 signalling pathway on the one hand, and by activating Foxl2 on the other hand.
Conclusions
Our model reproduces a wide range of experimental data reported for the development of wild type and mutant gonads. It also provides a formal support to crucial aspects of the gonad sexual development and predicts gonadal phenotypes for mutations not tested yet.

Drosophila DPP (TGF-beta homolog) signalling pathway is triggered by ligand-induced formation of heterotetrameric complexes consisting of two type II receptors and two type I receptors with intrinsic serine/threonine kinase activity. The type I receptor (SAX or TKV) is phosphorylated by the constitutively active type II receptor kinase (Punt). Consequently, the complex becomes active and phosphorylates the receptor-regulated Smads (R-Smads). Phosphorylated R-Smads (MAD and Smox) form complexes with a common-mediator Smad (Medea) and translocate into the nucleus, where they regulate the transcription of target genes in co-operation with other transcription factors (nejire, schnurri). DPP is a morphogen, i.e. a molecule distributed in a concentration gradient that elicits different cell fates as a function of its concentration, thereby organizing a series of cell types in a defined spatial array. In response to DPP gradient, cells adopt different fates. The establishment of dpp gradient involves the proteins SOG and TSG. These proteins together capture the DPP ligand and prevent its binding to the receptor (Punt). The heteromeric complex (SOG, DPP, TSG) then release the DPP ligand, a process involving the cleavage of SOG by Tolloid (a metalloprotease). Other TGF-beta signals, Glass-bottom-boa (GBB) and Screw (SCW), help DPP to potentiate cells to respond. SCW and GBB are never expressed together in the same region and affect different cells during: i) early D/V patterning of the embryo and specification/differentiation of dorsal cells (if there is no screw, dpp alone is unable to establish the D/V pattern and embryo lack amnioserosa); ii) the development of adult structures such as the wing. GBB or SCW form heterodimeric complexes with DPP. These heterodimers can only signal through TKV, while SCW/SCW and GBB/GBB signals trough SAX, and DPP/DPP trough TKV and SAX. To model DPP signalling and the formation of the gradient, we have considered three different levels for the TKV receptor (0, 1, 2) and the MADMED effector (0, 1, 2). The regulatory graph also accounts for the potentiation of responding cells due to association of DPP and SCW, or of DPP and GBB. Activated by MADMED, DAD is a pathway inhibitor that can modulate the pathway activity from high to low signalling. DAD works by abrogating the phosphorylation of the MADMED complex by TKV or SAX, thus involving a negative circuit between DAD and the MADMED complex. In addition, BRK another inhibitor of the DPP pathway can block the transcription of dad. Our model reproduces the formation of the DPP signalling gradient and accounts for the role of the heterodimers signalling in cell potentiation. To simulate DPP signalling, we start from an initial state corresponding to non differentiated cell, that can receives high or low level of DPP signal. The use of ternary nodes enables us to account for differential effects of different DPP levels (gradient). The cells receiving high level expression display the hetero-dimers SCW/DPP or GBB/DPP and correspond to Tld expression area, which promotes DPP gradient formation. In presence of medium DPP, TSG and SOG but no TLD are initially needed to capture homo- or hetero-dimer, diminishing pathway signalling intensity (expression level 1 for TKV and MADMED). In presence of high pathway signalling, two situations occur: i) in cells potentiated by SCW: a sequestering complex (SOG/TSG/ DPP/SCW) will release the signalling molecule upon TLD clivage, in addition to normal DPP signalling. This leads to a higher signal transduction. ii) in cells potentiated by GBB, the situation is similar but involve a different heterodimer (GBB/DPP). These situations correspond to two different stable states with high TKV and MADMED (level 2), denoting that more receptors are required to enable a higher level of nuclear MADMED. We consider five different initial states: i) the first one corresponds to the absence of signalling, i.e. absence of DPP; ii) the second one corresponds to medium signalling, characterized by the presence of Dpp at level 1 and of SCW; iii) the third one corresponds to medium signalling, characterized by the presence of Dpp at level 1 and of GBB); iv) the fourth one corresponds to the presence of DPP at level 2 and of SCW; v) the last one corresponds to the presence of DPP at level 2 and of GBB. These set of initial states enable the simulation of five situations. No signalling, two medium and two high signalling that characterize the behavior of the pathway. The stable state obtained with the no signalling simulation shows the absence of binding of the ligands to the receptors TKV and Punt (level of expression 0) and the non activation of target nodes. These medium signals simulations in presence of DPP, show the activation of the receptors (level of expression 1) and subsequent signalling cascade leading to the activation of pathway's targets. These medium signal are defined by the level of expression 1 for DPP, MADMED and TKV while in the high signalling sets, these nodes are expressed at level 2. The node Tkv is multi-valued because the high signalling is characterized by the binding of hetero dimers (DPP/SCW or DPP/GBB) signalling through TKV. Note that GBB and SCW don't have the same expression pattern. For more details on Dpp signalling pathway regulation see [1]; [2]; [3]; [4]; [5].

References

The Hedgehog (Hh) signalling pathway plays a crucial role in animal embryonic and organ development. In the wing imaginal disc of Drosophila melanogaster, Hh is induced and diffuses from the posterior compartment to activate the corresponding pathway in cells immediately anterior to the boundary. In these boundary cells, the Hh gradient induces target genes in distinct domains as a function of the Hh concentration. One Hh target is its own receptor Patched (Ptc), which sequesters Hh and impedes further diffusion, thereby refining the boundary.

We have delineated a multivalued logical model of the patterning process defining the cellular anterior-posterior boundary that includes the formation of Hh gradient and Hh signalling transduction. This model accounts for the activation of Hh target genes mediated by positive (CiA) and negative (CiR) regulatory products of the gene cubitus interruptus (ci). Wild-type and mutant simulations are carried out to assess the coherence of the model with experimental data and to obtain biological insights into this fundamental process.

In addition to recapitulating experimental data, our logical analysis leads to the delineation of three crucial features. First, CiA should be present in all boundary cells. Second, Ptc is regulated by CiA, but also tentatively by CiR. Third, the model predicts that Engrailed acts at different functional levels in boundary and posterior cells.