Limits...
Coordination logic of the sensing machinery in the transcriptional regulatory network of Escherichia coli.

Janga SC, Salgado H, Martínez-Antonio A, Collado-Vides J - Nucleic Acids Res. (2007)

Bottom Line: The active and inactive state of transcription factors in growing cells is usually directed by allosteric physicochemical signals or metabolites, which are in turn either produced in the cell or obtained from the environment by the activity of the products of effector genes.Finally we show that evolutionary families of TFs do not show a tendency to preserve their sensing abilities.Our results provide a detailed panorama of the topological structures of E. coli TRN and the way TFs they compose off, sense their surroundings by coordinating responses.

View Article: PubMed Central - PubMed

Affiliation: Programa de Genómica Computacional, Centro de Ciencias Genómicas, Universidad Nacional Autónoma de México, Cuernavaca, Morelos, 62100, México. sarath@ccg.unam.mx

ABSTRACT
The active and inactive state of transcription factors in growing cells is usually directed by allosteric physicochemical signals or metabolites, which are in turn either produced in the cell or obtained from the environment by the activity of the products of effector genes. To understand the regulatory dynamics and to improve our knowledge about how transcription factors (TFs) respond to endogenous and exogenous signals in the bacterial model, Escherichia coli, we previously proposed to classify TFs into external, internal and hybrid sensing classes depending on the source of their allosteric or equivalent metabolite. Here we analyze how a cell uses its topological structures in the context of sensing machinery and show that, while feed forward loops (FFLs) tightly integrate internal and external sensing TFs connecting TFs from different layers of the hierarchical transcriptional regulatory network (TRN), bifan motifs frequently connect TFs belonging to the same sensing class and could act as a bridge between TFs originating from the same level in the hierarchy. We observe that modules identified in the regulatory network of E. coli are heterogeneous in sensing context with a clear combination of internal and external sensing categories depending on the physiological role played by the module. We also note that propensity of two-component response regulators increases at promoters, as the number of TFs regulating a target operon increases. Finally we show that evolutionary families of TFs do not show a tendency to preserve their sensing abilities. Our results provide a detailed panorama of the topological structures of E. coli TRN and the way TFs they compose off, sense their surroundings by coordinating responses.

Show MeSH

Related in: MedlinePlus

Statistical significance of co-occurrence of sensing classes in Feed Forward Loop (FFL) motifs observed in the transcriptional network of E. coli. (a) In a FFL motif, which is a commonly seen topological structure in transcriptional regulatory networks, TF X regulates another TF Y and both jointly modulate the expression of the target gene Z. Hence TF X can be considered to be in the first position while TF Y can be thought to be in the second position. (b) Matrix shows statistical significance for occurrence of different sensing category combinations in FFLs using the z-scores calculated by comparing against 1000 sets of randomly generated FFLs as described in Materials and Methods section. Positive z-scores correspond to favored combinations of sensing classes in FFLs and vice versa. /z-scores/ >3.3 were considered significant as they corresponded to P-values <0.001, unless otherwise stated.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2175315&req=5

Figure 1: Statistical significance of co-occurrence of sensing classes in Feed Forward Loop (FFL) motifs observed in the transcriptional network of E. coli. (a) In a FFL motif, which is a commonly seen topological structure in transcriptional regulatory networks, TF X regulates another TF Y and both jointly modulate the expression of the target gene Z. Hence TF X can be considered to be in the first position while TF Y can be thought to be in the second position. (b) Matrix shows statistical significance for occurrence of different sensing category combinations in FFLs using the z-scores calculated by comparing against 1000 sets of randomly generated FFLs as described in Materials and Methods section. Positive z-scores correspond to favored combinations of sensing classes in FFLs and vice versa. /z-scores/ >3.3 were considered significant as they corresponded to P-values <0.001, unless otherwise stated.

Mentions: Motifs are sub-graphs which occur more frequently than expected by chance in networks. They have been first described in the TRN of E. coli and subsequently found in a variety of complex systems (12,26). FFL is a three-node subgraph and is one of the most abundantly found motif in all well characterized TRNs studied so far (14). This motif comprises of three genes: a regulator X, which regulates Y, and gene Z which is regulated by both X and Y (Figure 1a). Unlike bifan motifs (see below) FFLs are not symmetric for the positions of the two TFs comprising this motif as the first TF regulates two genes while the second regulates only one. To understand the organization of TFs in FFLs in the context of sensing classification we first identified the complete set of FFLs in the currently known TRN of E. coli (see Materials and Methods section). A total of 659 FFLs could be identified which could be associated to TFs with a classified category of sensing. To address the contributions and enrichment of different classes of sensing for the positions of the two TFs which comprise a FFL, we compared distributions seen to randomly generated FFLs (see Materials and Methods section). We found ISM (Internal Sensing Metabolites class) to be significantly enriched for the first position of the FFL while the sensing classes H (Hybrid; sensing transported and synthesized metabolites), ETC (External sensing Two-Components) and ETM (External sensing Transported Metabolites) are underrepresented in the first position (Table 1). On the other hand, the second position is overrepresented by TFs from H, ETC and ETM and as expected suppressed for TFs from ISM. Interestingly, the IDB class, comprising of nucleoid-associated proteins responsible for controlling DNA topology and nucleoid organization, were not found to show any significant preference for the first or the second position however they were found to occur in a combinatorial fashion in FFLs (see below and Figure 1b). We then addressed the enrichment for combinatorial control by sensing classes in the entire pool of FFLs. To study this we compared against 1000 randomly generated FFL sets, same in size as the observed set, by shuffling TFs between pairs of FFLs as described in detail in Materials and Methods section. Figure 1b shows the pair wise combinations of sensing classes that were significantly enriched for positions 1 and 2 in FFLs. It should be noted that individual tendencies for either positions need not be the same as the pair-wise combinations in FFLs. We observed a clear and strong preference for the following series of combinations: ISM-H (P < 0.001), ISM-ETM (P < 0.001), ISM-ETC (P < 0.001), IDB-IDB (P < 0.001) and H-H (P < 0.004), suggesting that the first position of the FFL is preferentially occupied by ISM TFs when the second position is occupied by Hybrid (H) or one of the external class of TFs (ETM or ETC). These observations clearly suggest a strong coordination between the internal and external classes of TFs in FFLs. It is also interesting to note from this heatmap that IDB TFs strongly co-regulate their target promoters with only other of their kind. Similarly H TFs were also found to show this tendency indicating that these classes act independently and form self-consistent local structures. On the other hand, several other combinations did not show any preference for combinatorial control. In particular, we found that the combinations ISM-ISM, IDB-ISM, H-ISM, ETM-ISM, H-IDB, ETM-IDB, ETC-IDB, ETM-H, ETC-H, H-ETM, H-ETC, ETC-ETM, IDB-ETC and H-ETC were significantly underrepresented (P < 0.001) for the first and second TF positions, reinforcing that, although internal TFs dominantly occupy the first position of the FFL, they do not control their promoters independently but rather in coordination with the help of the external TFs. These observations also suggest that TFs sensing external signals almost never control FFLs i.e. they are not in the first position, but are mostly under the control of internal sensing TFs. Interestingly, neither IDB nor H, which can sense signals of internal origin, control the core internal ISM TFs when the later takes the second position, suggesting that neither nucleoid associated nor hybrid TFs start a FFL in coordination to responses from other kinds of sensing classes. However, IDB and H TFs tend to co-ordinate with TFs of the same class. It can also be noted that H and ETC or H and ETM TFs never work together in a FFL which is likely due to the fact that all H TFs, by definition, can sense signals of both internal and external origin and hence do not need any explicit co-ordination with TFs that only sense external signals.Figure 1.


Coordination logic of the sensing machinery in the transcriptional regulatory network of Escherichia coli.

Janga SC, Salgado H, Martínez-Antonio A, Collado-Vides J - Nucleic Acids Res. (2007)

Statistical significance of co-occurrence of sensing classes in Feed Forward Loop (FFL) motifs observed in the transcriptional network of E. coli. (a) In a FFL motif, which is a commonly seen topological structure in transcriptional regulatory networks, TF X regulates another TF Y and both jointly modulate the expression of the target gene Z. Hence TF X can be considered to be in the first position while TF Y can be thought to be in the second position. (b) Matrix shows statistical significance for occurrence of different sensing category combinations in FFLs using the z-scores calculated by comparing against 1000 sets of randomly generated FFLs as described in Materials and Methods section. Positive z-scores correspond to favored combinations of sensing classes in FFLs and vice versa. /z-scores/ >3.3 were considered significant as they corresponded to P-values <0.001, unless otherwise stated.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2175315&req=5

Figure 1: Statistical significance of co-occurrence of sensing classes in Feed Forward Loop (FFL) motifs observed in the transcriptional network of E. coli. (a) In a FFL motif, which is a commonly seen topological structure in transcriptional regulatory networks, TF X regulates another TF Y and both jointly modulate the expression of the target gene Z. Hence TF X can be considered to be in the first position while TF Y can be thought to be in the second position. (b) Matrix shows statistical significance for occurrence of different sensing category combinations in FFLs using the z-scores calculated by comparing against 1000 sets of randomly generated FFLs as described in Materials and Methods section. Positive z-scores correspond to favored combinations of sensing classes in FFLs and vice versa. /z-scores/ >3.3 were considered significant as they corresponded to P-values <0.001, unless otherwise stated.
Mentions: Motifs are sub-graphs which occur more frequently than expected by chance in networks. They have been first described in the TRN of E. coli and subsequently found in a variety of complex systems (12,26). FFL is a three-node subgraph and is one of the most abundantly found motif in all well characterized TRNs studied so far (14). This motif comprises of three genes: a regulator X, which regulates Y, and gene Z which is regulated by both X and Y (Figure 1a). Unlike bifan motifs (see below) FFLs are not symmetric for the positions of the two TFs comprising this motif as the first TF regulates two genes while the second regulates only one. To understand the organization of TFs in FFLs in the context of sensing classification we first identified the complete set of FFLs in the currently known TRN of E. coli (see Materials and Methods section). A total of 659 FFLs could be identified which could be associated to TFs with a classified category of sensing. To address the contributions and enrichment of different classes of sensing for the positions of the two TFs which comprise a FFL, we compared distributions seen to randomly generated FFLs (see Materials and Methods section). We found ISM (Internal Sensing Metabolites class) to be significantly enriched for the first position of the FFL while the sensing classes H (Hybrid; sensing transported and synthesized metabolites), ETC (External sensing Two-Components) and ETM (External sensing Transported Metabolites) are underrepresented in the first position (Table 1). On the other hand, the second position is overrepresented by TFs from H, ETC and ETM and as expected suppressed for TFs from ISM. Interestingly, the IDB class, comprising of nucleoid-associated proteins responsible for controlling DNA topology and nucleoid organization, were not found to show any significant preference for the first or the second position however they were found to occur in a combinatorial fashion in FFLs (see below and Figure 1b). We then addressed the enrichment for combinatorial control by sensing classes in the entire pool of FFLs. To study this we compared against 1000 randomly generated FFL sets, same in size as the observed set, by shuffling TFs between pairs of FFLs as described in detail in Materials and Methods section. Figure 1b shows the pair wise combinations of sensing classes that were significantly enriched for positions 1 and 2 in FFLs. It should be noted that individual tendencies for either positions need not be the same as the pair-wise combinations in FFLs. We observed a clear and strong preference for the following series of combinations: ISM-H (P < 0.001), ISM-ETM (P < 0.001), ISM-ETC (P < 0.001), IDB-IDB (P < 0.001) and H-H (P < 0.004), suggesting that the first position of the FFL is preferentially occupied by ISM TFs when the second position is occupied by Hybrid (H) or one of the external class of TFs (ETM or ETC). These observations clearly suggest a strong coordination between the internal and external classes of TFs in FFLs. It is also interesting to note from this heatmap that IDB TFs strongly co-regulate their target promoters with only other of their kind. Similarly H TFs were also found to show this tendency indicating that these classes act independently and form self-consistent local structures. On the other hand, several other combinations did not show any preference for combinatorial control. In particular, we found that the combinations ISM-ISM, IDB-ISM, H-ISM, ETM-ISM, H-IDB, ETM-IDB, ETC-IDB, ETM-H, ETC-H, H-ETM, H-ETC, ETC-ETM, IDB-ETC and H-ETC were significantly underrepresented (P < 0.001) for the first and second TF positions, reinforcing that, although internal TFs dominantly occupy the first position of the FFL, they do not control their promoters independently but rather in coordination with the help of the external TFs. These observations also suggest that TFs sensing external signals almost never control FFLs i.e. they are not in the first position, but are mostly under the control of internal sensing TFs. Interestingly, neither IDB nor H, which can sense signals of internal origin, control the core internal ISM TFs when the later takes the second position, suggesting that neither nucleoid associated nor hybrid TFs start a FFL in coordination to responses from other kinds of sensing classes. However, IDB and H TFs tend to co-ordinate with TFs of the same class. It can also be noted that H and ETC or H and ETM TFs never work together in a FFL which is likely due to the fact that all H TFs, by definition, can sense signals of both internal and external origin and hence do not need any explicit co-ordination with TFs that only sense external signals.Figure 1.

Bottom Line: The active and inactive state of transcription factors in growing cells is usually directed by allosteric physicochemical signals or metabolites, which are in turn either produced in the cell or obtained from the environment by the activity of the products of effector genes.Finally we show that evolutionary families of TFs do not show a tendency to preserve their sensing abilities.Our results provide a detailed panorama of the topological structures of E. coli TRN and the way TFs they compose off, sense their surroundings by coordinating responses.

View Article: PubMed Central - PubMed

Affiliation: Programa de Genómica Computacional, Centro de Ciencias Genómicas, Universidad Nacional Autónoma de México, Cuernavaca, Morelos, 62100, México. sarath@ccg.unam.mx

ABSTRACT
The active and inactive state of transcription factors in growing cells is usually directed by allosteric physicochemical signals or metabolites, which are in turn either produced in the cell or obtained from the environment by the activity of the products of effector genes. To understand the regulatory dynamics and to improve our knowledge about how transcription factors (TFs) respond to endogenous and exogenous signals in the bacterial model, Escherichia coli, we previously proposed to classify TFs into external, internal and hybrid sensing classes depending on the source of their allosteric or equivalent metabolite. Here we analyze how a cell uses its topological structures in the context of sensing machinery and show that, while feed forward loops (FFLs) tightly integrate internal and external sensing TFs connecting TFs from different layers of the hierarchical transcriptional regulatory network (TRN), bifan motifs frequently connect TFs belonging to the same sensing class and could act as a bridge between TFs originating from the same level in the hierarchy. We observe that modules identified in the regulatory network of E. coli are heterogeneous in sensing context with a clear combination of internal and external sensing categories depending on the physiological role played by the module. We also note that propensity of two-component response regulators increases at promoters, as the number of TFs regulating a target operon increases. Finally we show that evolutionary families of TFs do not show a tendency to preserve their sensing abilities. Our results provide a detailed panorama of the topological structures of E. coli TRN and the way TFs they compose off, sense their surroundings by coordinating responses.

Show MeSH
Related in: MedlinePlus