Periodic Report Summary - NMSSBLS (Nonlinear mechanisms of spatial symmetry breaking in living systems)
Nonlinear network architecture shapes the dynamics and response of the transforming growth factor beta (TGF-ß) pathway
Introduction
The cellular machinery is governed by interacting proteins, genes and metabolites that form complex and highly interconnected networks of interactions. This way, extracellular stimuli triggers pathways of biological events that regulate gene expression, protein activity and ultimately, cell response. TGF-ß pathway is one of the most conserved and prolific of these signalling cascades, involved in a wide variety of both adult and embryonic processes. The TGF-ß pathway regulates growth, proliferation, differentiation and apoptosis; despite its quite simple architecture: extracellular TGF-ß and activin ligands bind to a type II transmembrane receptor, triggering the assembly of the tetrameric active ligand-receptor complex by binding to a type I receptor. Active complex then recruits and phosphorylates downstream effectors smad2 and smad3 (R-smads) on their C terminus residues. These smad proteins form hetero-trimeric complexes with smad4 that enter the nuclei and recruit cofactors to ultimately regulate gene expression. The deoxyribonucleic acid (DNA) binding complex can contain either two smad2 molecules and one smad 4, two smad3 and one smad4, but also one smad2, one smad3 and one smad 4. Despite their high similarity in sequence, i.e. 91% amino-acid sequence similarity, 66% between their MAD homology one (MH1) domain and 96% between their MH2 domains, the smad2 and smad3 have been shown to be functionally nonequivalent, recruiting different cofactors and targeting different regulatory sequences.
Many studies have focussed in comparing the roles of smad2 and smad3 following TGF-ß activation. Smad2 and smad3 often share a similar function, playing a redundant role in ovarian granulosa cells, in mesoderm induction in zebrafish and many other cellular contexts. More importantly, smad3 has been shown to rescue smad2 deletion phenotype in developing mouse embryo, when introduced into the smad2 locus. On the other hand, several systems show antagonistic behaviour of the two smads, with smad2 reducing the activity of smad3 and vice versa. For instance, smad and smad3 have been shown to have opposite roles in breast cancer bone metastasis by affecting differentially tumour angiogenesis. Smad2 and smad3 are antagonist in regulating growth and cell migration in pancreatic adenocarcinoma cells. Smad2 and smad3 also have opposite effects in TGF-ß dependent transcription through FAST, a fork-head DNA-binding protein. In addition, smad2 and smad3 have an opposite effect on the antioxidant response reporter (ARE). Transcriptional profiling experiments have shown smad3 activated genes that are repressed by smad2. The question of how such similar molecules can cooperate or antagonise each other, despite targeting different regulatory sequences is still open.
Our studies show that smad2 and smad3 cooperate and antagonise each other in the same cellular context, as a result of the particular wiring of the network of interactions of the TGF-ß pathway. In particular, the interaction between the two transcription factors forming the heterotrimeric complex smad2-smad3-smad4 after activation, bias the signal towards antagonism or cooperation between the smads. This positive or negative cooperation occurs despite the fact that the two molecules regulate different set of genes.
To show that, we proceed by first developing a mathematical model of the TGF-ß pathway to explore how the architecture of the pathway and the interaction between the smads induces cooperativity or antagonism. Then we test the model predictions in the context of neural development in an in vivo system. Our previous studies have shown how TGF-ß pathway activation and in particular smad3, promotes cell-cycle exit and neurogenesis by inhibiting the expression of Id proteins while activating the expression of neurogenic factors. We proceed by characterising the expression and function in neural development of smad2 and compare it with smad3. Next, we performed in ovo electroporation to induce overexpression and reduction of smad2 and smad3 and compare the effect on both smad2 and smad3 direct targets. Then, we tested the effect of gain and loos of function of each of the R-smads in neurogenesis. Rather than simply cooperativity or antagonism, the experiments reveal a complex scenario where both smads cooperate and antagonise, even in the same cellular context. We finally incorporate the experimental observations to the mathematical model to observe that all experimental results fit with a scenario where the complex smad2-smad3-smad4 plays a critical role in sequestering the smads after activation and enhancing the activity of smad3, but not smad2.
Our main results are:
Mathematical model of the pathway predicts that smad2 and smad3 can be antagonists or redundant depending on the hererotrimeric complex. To understand how cooperatively or antagonism between smad2 and smad3 arises, we develop a theoretical model that captures the basis of the TGF-ß pathway interactions. The model allows testing different hypotheses for the function of the heterotrimer complex formed by one molecule of smad2, smad3 and smad4.
Smad3 gets strongly activated caudally and inhibited in migratory cells. To analyse the in vivo activity of the TGF-ß pathway driven by smad2 and smad3, we took advantage of different reporter constructs driven by binding sequences specific for each of the smads. Our experiments have shown that smad3 shows an axial pattern of activation, with stronger activation in caudal sections. Smad3 activation is reduced in differentiated cells and migratory neural crest cells.
Gain of function of smad2 cooperates with smad3, while smad3 antagonises smad2. Next, we use the specific reporters for smad2 and smad3 to study the effect in smad3 specific targets after electroporation of smad2 and vice-versa. These results show that smad2 and smad3 crosstalk and influence each other specific targets, in such a way that smad2 and smad3 cooperate in smad3-direct targets but antagonise in smad2-direct targets. Moreover, pathway activation and formation of the smad2-smad3 transcriptional active complexes potentiates smad3-direct targets while reducing smad2-specific expression.
Smad2 and smad3 cooperate in neurogenesis. To test whether this cooperation has an effect in neurogenesis of smad2 in differentiation, we quantify the percentage of electroporated differentiated cells after electroporation of different preparations with same final concentration of hsmad2, hsmad3 and smad2 and smad3. Our data suggests a scenario of positive cooperation of smad2 with smad3 in neurogenesis.
Smad2 can also antagonise smad3 in smad3 specific transcription. Our previous experiments evidenced that exogenous smad2 and smad3 cooperate in smad3-specific readouts, but antagonise in smad2-specific readouts. To understand whether this effect is due to the formation of a combined heterotrimer smad2/3/4, we designed short-hairpin ribonucleic acid's (RNA's) highly specific to smad2 and smad3, restricting the transcriptional activity of the r-smads to the smad3/3/4 and smad2/2/4 complex, respectively.
The mathematical model predicts that the heterotrimer complex can activate smad3 targets but not smad2. We then rewrote the mathematical model developed in the first section to account for the experimental observations. The experiments fit with a scenario where smad2 and smad3 cooperate forming the heterotrimer complex that has an active role in activating mainly smad3 specific transcription. The heterotrimer formation induces the cooperation and antagonism between the r-smads.
Conclusions:
We have shown how naturally occurring nonlinear network motifs shape the response and function of proteins in key signalling cascades, determining the response of the cell to external stimuli and controlling key cellular aspects such as proliferation and differentiation. In particular, we focussed on the TGF-ß pathway, where we have shown numerically and experimentally how the nonlinear network motif created by the interaction between the R-smads to form the transcriptional complexes creates antagonism and cooperation between the two proteins.
Introduction
The cellular machinery is governed by interacting proteins, genes and metabolites that form complex and highly interconnected networks of interactions. This way, extracellular stimuli triggers pathways of biological events that regulate gene expression, protein activity and ultimately, cell response. TGF-ß pathway is one of the most conserved and prolific of these signalling cascades, involved in a wide variety of both adult and embryonic processes. The TGF-ß pathway regulates growth, proliferation, differentiation and apoptosis; despite its quite simple architecture: extracellular TGF-ß and activin ligands bind to a type II transmembrane receptor, triggering the assembly of the tetrameric active ligand-receptor complex by binding to a type I receptor. Active complex then recruits and phosphorylates downstream effectors smad2 and smad3 (R-smads) on their C terminus residues. These smad proteins form hetero-trimeric complexes with smad4 that enter the nuclei and recruit cofactors to ultimately regulate gene expression. The deoxyribonucleic acid (DNA) binding complex can contain either two smad2 molecules and one smad 4, two smad3 and one smad4, but also one smad2, one smad3 and one smad 4. Despite their high similarity in sequence, i.e. 91% amino-acid sequence similarity, 66% between their MAD homology one (MH1) domain and 96% between their MH2 domains, the smad2 and smad3 have been shown to be functionally nonequivalent, recruiting different cofactors and targeting different regulatory sequences.
Many studies have focussed in comparing the roles of smad2 and smad3 following TGF-ß activation. Smad2 and smad3 often share a similar function, playing a redundant role in ovarian granulosa cells, in mesoderm induction in zebrafish and many other cellular contexts. More importantly, smad3 has been shown to rescue smad2 deletion phenotype in developing mouse embryo, when introduced into the smad2 locus. On the other hand, several systems show antagonistic behaviour of the two smads, with smad2 reducing the activity of smad3 and vice versa. For instance, smad and smad3 have been shown to have opposite roles in breast cancer bone metastasis by affecting differentially tumour angiogenesis. Smad2 and smad3 are antagonist in regulating growth and cell migration in pancreatic adenocarcinoma cells. Smad2 and smad3 also have opposite effects in TGF-ß dependent transcription through FAST, a fork-head DNA-binding protein. In addition, smad2 and smad3 have an opposite effect on the antioxidant response reporter (ARE). Transcriptional profiling experiments have shown smad3 activated genes that are repressed by smad2. The question of how such similar molecules can cooperate or antagonise each other, despite targeting different regulatory sequences is still open.
Our studies show that smad2 and smad3 cooperate and antagonise each other in the same cellular context, as a result of the particular wiring of the network of interactions of the TGF-ß pathway. In particular, the interaction between the two transcription factors forming the heterotrimeric complex smad2-smad3-smad4 after activation, bias the signal towards antagonism or cooperation between the smads. This positive or negative cooperation occurs despite the fact that the two molecules regulate different set of genes.
To show that, we proceed by first developing a mathematical model of the TGF-ß pathway to explore how the architecture of the pathway and the interaction between the smads induces cooperativity or antagonism. Then we test the model predictions in the context of neural development in an in vivo system. Our previous studies have shown how TGF-ß pathway activation and in particular smad3, promotes cell-cycle exit and neurogenesis by inhibiting the expression of Id proteins while activating the expression of neurogenic factors. We proceed by characterising the expression and function in neural development of smad2 and compare it with smad3. Next, we performed in ovo electroporation to induce overexpression and reduction of smad2 and smad3 and compare the effect on both smad2 and smad3 direct targets. Then, we tested the effect of gain and loos of function of each of the R-smads in neurogenesis. Rather than simply cooperativity or antagonism, the experiments reveal a complex scenario where both smads cooperate and antagonise, even in the same cellular context. We finally incorporate the experimental observations to the mathematical model to observe that all experimental results fit with a scenario where the complex smad2-smad3-smad4 plays a critical role in sequestering the smads after activation and enhancing the activity of smad3, but not smad2.
Our main results are:
Mathematical model of the pathway predicts that smad2 and smad3 can be antagonists or redundant depending on the hererotrimeric complex. To understand how cooperatively or antagonism between smad2 and smad3 arises, we develop a theoretical model that captures the basis of the TGF-ß pathway interactions. The model allows testing different hypotheses for the function of the heterotrimer complex formed by one molecule of smad2, smad3 and smad4.
Smad3 gets strongly activated caudally and inhibited in migratory cells. To analyse the in vivo activity of the TGF-ß pathway driven by smad2 and smad3, we took advantage of different reporter constructs driven by binding sequences specific for each of the smads. Our experiments have shown that smad3 shows an axial pattern of activation, with stronger activation in caudal sections. Smad3 activation is reduced in differentiated cells and migratory neural crest cells.
Gain of function of smad2 cooperates with smad3, while smad3 antagonises smad2. Next, we use the specific reporters for smad2 and smad3 to study the effect in smad3 specific targets after electroporation of smad2 and vice-versa. These results show that smad2 and smad3 crosstalk and influence each other specific targets, in such a way that smad2 and smad3 cooperate in smad3-direct targets but antagonise in smad2-direct targets. Moreover, pathway activation and formation of the smad2-smad3 transcriptional active complexes potentiates smad3-direct targets while reducing smad2-specific expression.
Smad2 and smad3 cooperate in neurogenesis. To test whether this cooperation has an effect in neurogenesis of smad2 in differentiation, we quantify the percentage of electroporated differentiated cells after electroporation of different preparations with same final concentration of hsmad2, hsmad3 and smad2 and smad3. Our data suggests a scenario of positive cooperation of smad2 with smad3 in neurogenesis.
Smad2 can also antagonise smad3 in smad3 specific transcription. Our previous experiments evidenced that exogenous smad2 and smad3 cooperate in smad3-specific readouts, but antagonise in smad2-specific readouts. To understand whether this effect is due to the formation of a combined heterotrimer smad2/3/4, we designed short-hairpin ribonucleic acid's (RNA's) highly specific to smad2 and smad3, restricting the transcriptional activity of the r-smads to the smad3/3/4 and smad2/2/4 complex, respectively.
The mathematical model predicts that the heterotrimer complex can activate smad3 targets but not smad2. We then rewrote the mathematical model developed in the first section to account for the experimental observations. The experiments fit with a scenario where smad2 and smad3 cooperate forming the heterotrimer complex that has an active role in activating mainly smad3 specific transcription. The heterotrimer formation induces the cooperation and antagonism between the r-smads.
Conclusions:
We have shown how naturally occurring nonlinear network motifs shape the response and function of proteins in key signalling cascades, determining the response of the cell to external stimuli and controlling key cellular aspects such as proliferation and differentiation. In particular, we focussed on the TGF-ß pathway, where we have shown numerically and experimentally how the nonlinear network motif created by the interaction between the R-smads to form the transcriptional complexes creates antagonism and cooperation between the two proteins.