The Regulatory Network of Pseudomonas aeruginosa
© Galán-Vásquez et al; licensee BioMed Central Ltd. 2011
Received: 15 December 2010
Accepted: 14 June 2011
Published: 14 June 2011
Pseudomonas aeruginosa is an important bacterial model due to its metabolic and pathogenic abilities, which allow it to interact and colonize a wide range of hosts, including plants and animals. In this work we compile and analyze the structure and organization of an experimentally supported regulatory network in this bacterium.
The regulatory network consists of 690 genes and 1020 regulatory interactions between their products (12% of total genes: 54% sigma and 16% of transcription factors). This complex interplay makes the third largest regulatory network of those reported in bacteria. The entire network is enriched for activating interactions and, peculiarly, self-activation seems to occur more prominent for transcription factors (TFs), which contrasts with other biological networks where self-repression is dominant. The network contains a giant component of 650 genes organized into 11 hierarchies, encompassing important biological processes, such as, biofilms formation, production of exopolysaccharide alginate and several virulence factors, and of the so-called quorum sensing regulons.
The study of gene regulation in P. aeruginosa is biased towards pathogenesis and virulence processes, all of which are interconnected. The network shows power-law distribution -input degree -, and we identified the top ten global regulators, six two-element cycles, the longest paths have ten steps, six biological modules and the main motifs containing three and four elements. We think this work can provide insights for the design of further studies to cover the many gaps in knowledge of this important bacterial model, and for the design of systems strategies to combat this bacterium.
Pseudomonas aeruginosa is a metabolically versatile Gram-negative bacterium, able to express a wide variety of virulence factors. These allow P. aeruginosa to grow in soil and marine habitats, as well as on plant and animal tissues. It is also a significant source of bacteraemia in burn victims, urinary-tract infections, hospital-acquired pneumonia and predominant cause of morbidity and mortality in cystic fibrosis patients . All these makes P. aeruginosa the most studied bacterial model regarding the control of pathogenic determinants and the third bacterial model more studied with respect to their molecular biology -after Escherichia coli and Bacillus subtilis-. The genome sequence of P. aeruginosa strain PAO1 was reported in 2000 , and since then numerous databases and genomic resources have been implemented to study their molecular and pathogenic biology [2–4].
It is well know the importance of gene regulation on the organisms' performance as this process defines their metabolic, adaptive and pathogenic capabilities. In this work, we report a collection of known regulatory network interactions connecting transcription factors (TFs), sigma factors (σ), and anti-sigma factors to their target genes in P. aeruginosa. This transcriptional regulatory network (TRN) constitutes the third largest one of any bacteria reported to date. We proceed to analyze the main topological properties of this network and the main functional interactions among their regulatory components. We hope these results will provide insights and guide future studies to increase our knowledge on this important bacterium.
Results and discussion
The transcriptional regulatory network (TRN) of Pseudomonas aeruginosa
Topological description of the TRN in P. aeruginosa
The clustering coefficient C, is a measure that indicates the probability that two genes with a common neighbor in a graph are also interconnected; that is to say, the clustering coefficient quantifies what so much the local neighborhood of a gene is as member of a group of genes. It is common for networks to exhibit a decreasing value of C(k) with respect to the degree k, such that in small groups or modules of genes the elements are well connected, but as the group increases in size the elements are progressively less connected. The regulatory network of P. aeruginosa shares this general clustering property in (Figure 3D).
Connectivity in a network refers to the associations between every pair of genes.
Connections can be via a direct link or indirectly through a series of intermediate interactions. Connected components are defined for undirected networks, and give us information about how much are connected the elements in a network and their modular structure. Sometimes it is necessary to consider the network as undirected, since it allows us to capture different types of information to perform a better analysis. In the case of the TRN in P. aeruginosa there are 12 connected components, with one giant component containing 650 genes, while the rest contain at most six genes. Each connected component in the TRN possesses at least one TF or σ. A skeleton of 65 TFs and 13 σ maintain cohesive this giant component (the 12% of its components). We consider that a connected component is composed by n nodes, and calculate the relative frequency P(n) for every possible n, which give us the distribution of the number of nodes in a connected component (Figure 3E), [see also Additional file 1].
Functional organization of the regulatory network
In order to discern the functional organization of a regulatory network we can study the following aspects of the TRN: i) the regulatory mode and connectivity of each component of the transcriptional machinery and, ii) the manner in which endogenous and exogenous information, relevant for transcriptional regulation, enter and pass through the regulatory machinery until conclude on promoters of target genes. All these computes should be associated with the biological functions of the respective genes. In this sense, some interesting findings in the TRN of P. aeruginosa are discussed below.
Activation is the dominating activity in the TRN
Number of TFs
Average path length
Maximum out degree
Maximum in degree
Most of the TFs are positively auto-regulated
Top ten most influencing regulators in the TRN of P. aeruginosa
TFs and σ regulated (excluding self-regulation)
Total of genes regulated
Type of σ used by the regulated promoters
Number of TFs used as co-regulators
Short paths are common in metabolic and signal transduction networks, since this arrangement ensures fast and efficient response to changes in food use and to environmental perturbations . The longest paths in the TRN of E. coli include regulatory process for biofilm formation and flagella assembly , both of these are considered development processes and are also amongst the longest paths in P. aeruginosa.
A motif in a TRN is a topological structure that is more frequented than expected . The most represented motifs in the P. aeruginosa network are those formed by three and four genes (Figure 6), [Additional file 1]. Previous research suggests that motifs represent elements for optimal network design given their relationship with the network dynamics and structural stability. The prevalence of certain types of motifs has been considered a product of the evolution acting on the organization of biological networks [19–21]. In particular, motifs such as feed-forward loops (FFL; networks with three vertex, composed of two input transcription factors, one of which regulates the other, both jointly regulating a target gene) have a higher abundance in TRN than expected from random networks with the same number of nodes and arrows [18, 22]. The dynamic behavior of feed-forward loops has been extensively analyzed ; these studies revealed that FFL have two main functions: a) to speed up the response time of the target gene (incoherent FFL, when the signs of the direct and indirect regulation are opposites) and, b) to act as sign-sensitive delays for one of the two TFs (coherent FFL, with the same sign for both the direct and indirect regulation). Considering all the biological process where they participate, FFL are also implicated in pulse generation and cooperativity. In P. aeruginosa the most common motifs are those of three nodes known as coherent feed-forward loops , where the sign of the interactions is the same, positive in this case (Figure 6B). This type of motif is present 89 times in the P. aeruginosa regulatory network. Additionally, we found that the most common motifs of four nodes, which occurred 3832 in the network, are those known as bi-fan, where two TFs each positively co-regulate to two target genes (Figure 6C). This motif is also frequent in other organisms such as S. cerevisiae and E. coli .
Hierarchical organization of the TRN of P. aeruginosa
A hierarchical organization is given by a directed informational flux, beginning from the most influencing regulators. In this way, the TFs constitute the skeleton and the non-regulatory genes are the leaves in a hierarchical network (Figure 2 andAdditional file 1). The first level is populated by 33 TFs and 2 sigma factors. The origons , which are the points of informational inputs into the network, are set at this level. The second level is the most populated level but includes a high proportion of non-regulatory genes. Most of σ are set at higher levels, except for those involved in iron metabolism, which, as also observed for E. coli, are in the lowest levels as dedicated sigma factors for specific functions.
Promoter and regulatory regions are zones in transcription units where regulatory information is integrated. This regulatory integration is evidenced by the presence of DNA-binding sites for multiple regulators or promoters for different σ respectively. In the TRN of P. aeruginosa, regulators of level one can co-regulate with regulators of any other level and this co-regulatory activity diminishes as TFs are set lower in the network hierarchy. Regulators that frequently co-regulate with other regulators (in parenthesis the number of co-regulations) are: lasR(20), mexT(15), algR(14), ihf(14), rhlR(13), anr(12), ptxR(12). On the other hand, the most regulated genes are (in parenthesis the number of TFs regulating it): rhlI(8), rhlAR(7), algD(7), alg44(6), alg8(6), algA(6), algE(6) algF(6), algG(6), algI(6), algJ(6), algK(6), algL(6), algX(6), hcnA(6), lasR(6), algU(6). It is interesting that the most regulated genes in both E. coli and P. aeruginosa encode for TFs, FhlCD (flagella synthesis) and RhlI (quorum sensing) respectively.
Most influencing regulators in the TRN of P. aeruginosa
The most influencing regulators in a regulatory network are called "global regulators" and are defined by a series of operative properties, including: i) they should regulate a large number of genes; ii) should regulate other sigmas and regulators; iii) should co-regulate together with many TFs and, iv) their target genes should have promoters using more than one kind of σ . All these properties were computed for regulators found in P. aeruginosa (see Methods section) and the top ten are shown in Table 2. A coefficient G was introduced here, which indicates if a regulator is more or less global taking into account the regulatory criteria mentioned above. The most influencing regulators in Pseudomonas have a lower qualification than the corresponding seven global TFs in E. coli. This might be due to the limited knowledge of the transcriptional regulation in P. aeruginosa compared to E. coli.
Biological processes in P. aeruginosa TRN
Defining functional modules in a formal computational way is a difficult task. However, it has been shown that employing a simple metric of shorter distances among TFs in the E. coli network, it might be possible to recover modules with a good approximation to those that are manually defined, on the basis of the knowledge of biological functions of their products . In this work we used this metric for the TFs and σ -anti-σ sub-network of P. aeruginosa (Figure 4), and get the following biological modules: alginate biosynthesis, quorum sensing, iron capture and metabolism, production of virulence factors, antibiotic resistance and motility (Figure 5A). This finding was corroborated by manual inspection of TFs participating in the same biological processes (Figure 5B). It is clear that the processes that are more thoroughly studied in P. aeruginosa correspond to those related to pathogenesis and virulence properties while little attention has been given to biological processes, such as, central metabolism, membrane biogenesis, cell-division, etc. Most of the best-studied biological processes are connected, beginning from alginate biosynthesis to quorum sensing, and from there to those involved in the production of virulence factors. Additionally, there is a directed regulatory connection from alginate biosynthesis to iron metabolism and to some mechanisms of antibiotic resistance (Figure 5B). Since these processes act cooperatively during infection and pathogenesis, it is very important to give a detailed characterization of P. aeruginosa regulatory network. The latter may lead to the development of strategies to disrupt its connectivity, thus, possibly decreasing the pathogenicity of this bacterium.
Here we report the topological and functional organization of the third largest regulatory network in bacteria. From our analysis, it is evident that the study of regulation in P. aeruginosa is biased towards particular biological processes, involved in pathogenesis and virulence. These processes include alginate and biofilm formation, production of virulence factors and antibiotic resistance, many of which are coordinated by quorum sensing in the bacterial population. Current data suggests, that motility, iron metabolism and anaerobic respiration might be less connected to these processes by now. All these processes are connected in the network via a hierarchical organization with 11 levels, and the connected parts of the network form a giant component with 650 genes, the 10% of them corresponded to TFs. Overall, the network has degree distribution and structural organizations as other biological networks known to date. A peculiar property of this network is the fact that its TFs are mainly auto-activated. This is the first time this mode of self-regulation is reported as dominant in a bacterial TRN. It remains to be revealed whether this property is really a characteristic of the entire network in this bacterium, or is just is property of this part of the network, which clearly controls adaptive, pathogenic and virulence processes. As it can be observed, regulatory information related to several important biological processes of P. aeruginosa is lacking; for instance, the regulation on the uptake of carbon sources and their metabolism, amino acid biosynthesis or cell-division. This bias makes difficult a complete analysis on the regulatory network of this bacterium and better compare it with regulatory networks of bacteria most characterized such as E. coli or B. subtilis. It might be that studying basic biological functions on this organisms we can understand the basis of their versatile metabolism, adaptiveness and pathogenecity. In special it is lacking the knowledge of the activity of the housekeeping sigma and transcription factors controlling activities of central metabolism. Because of this it will be very important for the community working on the biology of P. aeruginosa to study additional biological processes in order to have a more complete picture of the regulatory network in this bacterium. We hope this analysis will give insights in this direction to guide future work, with the aim of covering the many gaps of knowledge on this important bacterial model.
Biological data and representation
The general strategy for the curation of regulatory interactions is shown in Figure 1. Briefly, we searched PubMed with relevant key words, such as: P. aeruginosa, sigma or transcription factor, transcriptional regulation, etc. Data on regulatory networks were obtained from the literature and compiled in an Excel table including experimental evidence and references. The Additional file 1 shows the complete information for the interactions of the entire network. The regulatory interactions were drawn in a form of network using the Cytoscape software .
Transcriptional machinery sub-network
With the aim of analyzing the regulatory behavior of the transcriptional machinery of P. aeruginosa, the regulatory interactions present only among TFs, sigmas and anti-sigmas from the whole network were extracted (Figure 4).
Computational analysis of the regulatory network
All the computational analyses on the network were made using the Octave free software http://www.octave.org. Analyses of degree, centrality, clustering coefficient, connectivity, cycles, paths and hierarchical levels were made according to previous definitions and following the approach as in . Motif determination was made following the work by Uri Alon and coworkers calculating the probability of finding the same motif in a random network as the average of the motifs found in 1000 randomized networks, maintaining the same number of nodes, edges and the proportion of the type of regulatory interactions (positive, negative, dual) .
The G coefficient for global regulators
Where, N TF indicates the total number of TFs (in the known network in each case), N G is the number of non-regulatory genes, and N SF is the number of sigma factors in the whole network. Additionally, TFR and GR, represents the number of TFs and non-regulatory genes regulated by each TF, respectively; SF represents the distinct sigma factors used by the promoters of genes regulated by each TF; and CR represents the number of TFs each TF co-regulates with.
Determination of biological modules/processes in the regulatory machinery network
With the aim of determining biological modules in the TRN we used a shortest path metric criteria among TFs and sigmas (we used the relation 1/D 2 where D is the distance between two nodes), as reported for E. coli . Additionally, we manually grouped TFs and sigma factors in agreement to the functional classification of their regulated genes .
List of abbreviations
transcriptional regulatory network.
Authors thank Angela Porras Dorantes and Viridiana Sanchez Paez for their help on compiling original literature, Johannes Klein for supplying some regulatory data from the Prodoric database and to Heladia Salgado for data management. We also thank Cei Abreu-Goodger, Gloria Soberón-Chávez, Carlos Cervantes, Luis Delaye and Jesús Campos for their useful comments on of the manuscript. EG-V received a CONACYT fellowship (234827) for a master's degree. This work was funded by young-researcher grants from CONCYTEG and CONACYT (102854) given to AM-A.
- Stover CK, Pham XQ, Erwin AL, Mizoguchi SD, Warrener P, Hickey MJ, Brinkman FS, Hufnagle WO, Kowalik DJ, Lagrou M, Garber RL, Goltry L, Tolentino E, Westbrock-Wadman S, Yuan Y, Brody LL, Coulter SN, Folger KR, Kas A, Larbig K, Lim R, Smith K, Spencer D, Wong GK, Wu Z, Paulsen IT, Reizer J, Saier MH, Hancock RE, Lory S, Olson MV: Complete genome sequence of Pseudomonas aeruginosa PAO1, an opportunistic pathogen. Nature. 2000, 406: 959-964. 10.1038/35023079.View ArticlePubMedGoogle Scholar
- Klein J, Leupold S, Münch R, Pommerenke C, Johl T, Kärst U, Jänsch L, Jahn D, Retter I: ProdoNet: identification and visualization of prokaryotic gene regulatory and metabolic networks. Nucleic Acids Res. 2008, 36: W460-W464. 10.1093/nar/gkn217.PubMed CentralView ArticlePubMedGoogle Scholar
- Choi C, Münch R, Leupold S, Klein J, Siegel I, Thielen B, Benkert B, Kucklick M, Schobert M, Barthelmes J, Ebeling C, Haddad I, Scheer M, Grote A, Hiller K, Bunk B, Schreiber K, Retter I, Schomburg D, Jahn D: SYSTOMONAS- an integrated database for systems biology analysis of Pseudomonas. Nucleic Acids Res. 2007, 35: D533-D537. 10.1093/nar/gkl823.PubMed CentralView ArticlePubMedGoogle Scholar
- Winsor G, Van T, Lo R, Bhavjinder K, Whiteside M, Hancock R, Brinkman S: Pseudomonas Genome Database: facilitating user-friendly, comprehensive comparisons of microbial genomes. Nucleic Acids Res. 2009, 37: D483-D488. 10.1093/nar/gkn861.PubMed CentralView ArticlePubMedGoogle Scholar
- Perez-Rueda E, Janga SC, Martinez-Antonio A: Scaling relationship on the gene content of transcriptional machinery in bacteria. Mol. BioSyst. 2009, 12: 494-501.Google Scholar
- Réka A: Scale-free networks in cell biology. J. Cell Sci. 2005, 118: 4947-4957. 10.1242/jcs.02714.View ArticleGoogle Scholar
- Watts J: The "new" science of networks. Annu. Rev. Sociol. 2004, 30: 243-270. 10.1146/annurev.soc.30.020404.104342.View ArticleGoogle Scholar
- Erdõs P, Renyi A: On random graphs I. Publ. Math. 1959, 6: 290-297.Google Scholar
- Barabási A, Oltvai Z: Networks biology: understanding the cell's functional organization. Nat. Rev. Genet. 2004, 5: 101-113. 10.1038/nrg1272.View ArticlePubMedGoogle Scholar
- Martinez-Antonio A, Janga SC, Thieffry D: Functional organisation of Escherichia coli transcriptional regulatory network. J. Mol. Biol. 2008, 381: 238-247. 10.1016/j.jmb.2008.05.054.PubMed CentralView ArticlePubMedGoogle Scholar
- Tagkopoulos I, Liu YC, Tavazoie S: Predictive behavior within microbial genetic networks. Science. 2008, 320: 1313-7. 10.1126/science.1154456.PubMed CentralView ArticlePubMedGoogle Scholar
- Amir M, Romano G, Groisman B, Yona A, Dekel E, Kupiec M, Dahan O, Pikpel Y: Adaptive prediction of environmental changes by microorganisms. Nature. 2009, 460: 220-225. 10.1038/nature08112.View ArticleGoogle Scholar
- Thomas R: Boolean formalization of genetic control circuits. J. Theor. Biol. 1973, 42: 563-585. 10.1016/0022-5193(73)90247-6.View ArticlePubMedGoogle Scholar
- Lagomarsino M, Jona P, Bassetti B, Isambert H: Hierarchy and feedback in the evolution of the Escherichia coli transcription network. Proc. Natl. Acad. Sci. USA. 2007, 104: 5516-5520. 10.1073/pnas.0609023104.View ArticleGoogle Scholar
- Alon U: Network motifs: theory and experimental approaches. Nat. Rev. Genet. 2007, 8: 450-461. 10.1038/nrg2102.View ArticlePubMedGoogle Scholar
- Moreno-Hagelsieb G, Latimer K: Choosing BLAST options for batter detection of orthologs as reciprocal best hits. Bioinformatics. 2008, 319-24. 24
- Watts D, Strogatz S: Collective dynamics of 'small-world' networks. Nature. 1998, 393: 440-442. 10.1038/30918.View ArticlePubMedGoogle Scholar
- Shen-Orr S, Milo R, Mangan S, Alon U: Network motifs in the transcriptional regulation network of Escherichia coli. Nature Genet. 2002, 31: 64-68. 10.1038/ng881.View ArticlePubMedGoogle Scholar
- Doyle J, Csete M: Motifs, control, and stability. PLoS Biol. 2005, 3 (11): e392-10.1371/journal.pbio.0030392.PubMed CentralView ArticlePubMedGoogle Scholar
- Prill RJ, Iglesias PA, Levchenko A: Dynamic properties of network motifs contribute to biological network organization. PLoS Biol. 2005, 3 (11): e343-10.1371/journal.pbio.0030343.PubMed CentralView ArticlePubMedGoogle Scholar
- Mangan S, Zaslaver A, Alon U: The coherent feedforward loop serves as a signsensitive delay element in transcription networks. J. Mol. Biol. 2003, 334: 197-204. 10.1016/j.jmb.2003.09.049.View ArticlePubMedGoogle Scholar
- Balázsi G, Barabási AL, Oltvai ZN: Topological units of environmental signal processing in the transcriptional regulatory network of Escherichia coli. Proc. Natl. Acad. Sci. USA. 2005, 102: 7841-7846. 10.1073/pnas.0500365102.PubMed CentralView ArticlePubMedGoogle Scholar
- Mangan S, Alon U: Structure and function of the feed-forward loop network motif. Proc. Natl. Acad. Sci. USA. 2003, 100: 11980-11985. 10.1073/pnas.2133841100.PubMed CentralView ArticlePubMedGoogle Scholar
- Milo R, Shen-Orr S, Itzkovitz S, Kashtan N, Chklovskii D, Alon U: Network Motifs: Simple building blocks of complex networks. Science. 2002, 298: 824-827. 10.1126/science.298.5594.824.View ArticlePubMedGoogle Scholar
- Balazsa G, Babarasi A L, Oltvai ZN: Topological units of enviromental signal processing in the transcriptional regulatory network of Escherichia coli. Proc. Natl. Acad. Sci. USA. 2005, 102: 7841-7846. 10.1073/pnas.0500365102.View ArticleGoogle Scholar
- Martinez-Antonio A, Collado-Vides J: Identifying global regulators in transcriptional regulatory networks in bacteria. Curr. Opin. Microbiol. 2003, 6: 482-489. 10.1016/j.mib.2003.09.002.View ArticlePubMedGoogle Scholar
- Resendis-Antonio O, Freyre-González J, Menchaca-Méndez R, Gutiérrez-Ríos R, Martínez Antonio A, Ávila-Sánchez C, Collado-Vides J: Molecular analysis of the transcriptional regulatory network of Escherichia coli. Trends Genet. 2005, 21 (1): 16-20. 10.1016/j.tig.2004.11.010.View ArticlePubMedGoogle Scholar
- Shannon P, Markiel A, Ozier O, Baliga N, Wang J, Ramage D, Amin N, Schwikowski B, Ideker T: Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003, 13: 2498-2504. 10.1101/gr.1239303.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.