Interest in 'pathway biology' has never been greater as we struggle to comprehend cellular systems from a combination of targeted studies and the deluge of data flowing from 'omics platforms. This is reflected by the escalating efforts to assemble pathway diagrams [30
], develop standards for depicting pathways [26
], software to support their construction [82
] and exchange [85
], and the development of approaches to model and predict pathway behaviour [13
]. Whilst arguably there is no such thing as a pathway only one big integrated network of molecular interactions, it is still useful to think in terms of pathways as being connected modules of this network. As such a pathway may be considered to consist of a specific biological input or event that initiates a series of directional interactions between the components of a system leading to an appropriate shift in cellular activity. As we begin to appreciate the potential complexity of these molecular networks, there is increasing interest in modelling pathways in order to expand our understanding of biology from the traditional gene-centric view of life, to a systems level appreciation of biological function.
The pathways described here are of central importance to understanding macrophage biology and therefore innate immunity, and the diagram provides a consensus view of these systems. This is not to say that the model is either viewed as complete or necessarily even correct, but only as a working model. As such it has been designed with the idea that it will need to be modified and expanded based on new publications, experimental observations or deeper insight into specific systems. All of the pathways depicted are reasonably well characterized and as such there is a relative abundance of information on them from a wide variety of sources. What we lacked prior to this work was a way of collating our understanding of these pathways and integrating this view with the abundance of data generated on these cells by ourselves and others. Our aim has therefore been the creation of a pathway diagram that graphically reflects the current view of a pathway system in a visibly intuitive manner. In so doing we wished to create a resource for data integration, pathway modelling and hypothesis generation. In order to achieve these objectives we found it necessary to modify both the PDN [26
] and EPN schemes [28
] for pathway depiction. Our original diagram [30
] acted as our framework for the current effort helping to highlight the many gaps in our understanding and together with developing interests in macrophage biology, helped to prioritize areas for future modelling. Modelling of the pathways continued to be based on labour-intensive curation of the literature. Post-graduate students were given an area of biology to examine, and all the resources for researching the literature and depicting their chosen pathway module. Regular debates on the progress of the pathway models were held, and through this process deficiencies were plugged in the graphical depiction of events, pathway content, notation, component labelling and the recording of the supporting information; a process which in itself was highly informative. An important point is that the diagrams can be shared and understood by all those familiar with notation, and as a result all the work presented here has been subjected to form of internal critiquing. However, each new area of biology included in the current diagram has presented its own problems in layout and concept representation. As a result there has been subtle but almost constant re-evaluation of various aspects of the notation scheme and as we have dealt with new issues in the depiction of different systems. We are now satisfied that mEPN scheme has matured to the point where we foresee little need to change the majority of the notation scheme presented here (see Freeman et al.
, 2010 [31
] and http://www.mepn-pathway.org
), although clearly the modelling of other systems and ideas from others may present a case for further modifications.
Pathway diagrams are a well established tool in our effort to interpret and explain results from functional genomics investigations. Overlay of results, usually from studies of the difference between one biological state and another, on top of pathway diagrams allows the investigator to visualize and link observations to defined pathways. BioLayout Express3D
, a network analysis tool developed by us, provides a powerful approach to visualize and analyze 'omics data from a variety of sources [77
]. We have recently implemented the import of .graphml files into BioLayout Express3D
and the tool now supports the visualization of pathway diagrams as 3D or 2D networks [76
]. A parser automatically converts the mEPN notation into the equivalent 3D notation scheme and can use the diagrams original node co-ordinates to layout the pathway. We have now also implemented the ability to export analyses from one dataset e.g. clustering of microarray gene expression data and import and overlay these analyses on to another network. As we have used standard gene nomenclature in the assembly of this pathway it is possible to map directly between gene identifiers from data to genes/proteins in the pathway. Analysis of the transcriptional response of mouse bone marrow derived macrophages (BMDM) to Ifn-β stimulation has been used here as an example (Figure and additional file 5
). In practice any number of lists with annotations can be imported as class-sets onto the pathway and one can envisage how this would facilitate the comparison of numerous data sets in the context of the macrophage pathway. Although the concept of data mapping onto pathways is not new and is supported by other pathway resources [6
] these pathways suffer from a number of issues pertaining to the lack standard graphical notation used to depict them. Furthermore the nature of the pathway presented here (in terms of scale, detail, formalised notation, range of pathways covered and integrated nature of their presentation) presents a valuable additional resource for those interested in macrophage biology or any of the pathways covered. Clearly the better and more extensive the pathway diagrams are the easier it will be to provide a working hypothesis on the interpretation of data. Increasingly, it is now experimental data that is helping to refine existing pathway models and observations that we do not understand that are driving our current modelling efforts.
As a note of interest on pathway topology, it has been suggested that many metabolic and signalling pathways, such as the TLR system, possess a 'bow-tie' or 'hourglass' structure [31
]. This is to say that the results from these modelling efforts where pathways are depicted using other notation systems or depicted based only on protein-protein interactions [94
], suggest that many signalling pathways possess structures where numerous inputs (ligand/receptor interactions) feed through a small number of 'hub' adaptor molecules to give rise to large and overlapping responses. In this view of signalling pathways proteins common to numerous different signalling systems e.g. MYD88 in the TLR system, are generally viewed as more important than other components in ensuring a network's robustness to perturbation due to their high degree of connectivity. Our model of the TLR system would also support the idea that many of the TLR's all signal through use of a limited number of proteins e.g. MYD88, IRAK4, TIFA, TRAF6, TOLLIP. In each case these proteins act as members of receptor-adaptor complexes propagating the signal induced by ligand binding. In the case of each TLR, the model suggests that further progression of the signal is dependent on a number of MAP3K7 complexes which subsequently go on to activate IKK complex and ultimately NF-κB transcription factors. However, there is also clear evidence linking TLR signalling to the activation of MAPK14 (p38) and MAPK1/3 (ERK) cascades [95
]. The literature would also suggest that engagement of endosomal TLRs and the MYD88-dependent TLR4 pathway leads to an activation of IFNB
expression via IRF3 [97
]. Hence the TLR system acts through numerous signalling networks to bring about the extensive transcriptional changes that are associated with immune activation by LPS or other TLR ligands. The signalling systems used by TLRs are also used by other receptors. For example, the activation of IFNB
expression via IRF3 is also utilized by ZBP1, DDX58 (RIG-I) and IFIH1 (MDA5) cytosolic DNA/RNA receptors [52
]. Activation of IFNB
undoubtedly contributes to the transcriptional signature observed when these systems are activated. Further complexity still is evident if we consider that these or indeed other immune responses are all clearly modulated by feedback control [102
]. Indeed, a number of negative regulators are observed to be some of the earliest up-regulated genes in the type 1 interferon response measured here e.g. Socs1/3
. At the same time many feed-forward loops are being established by the up-regulation of transcriptional regulators which go on to activate genes associated with effector systems/pathway modules. Our integrated model of macrophage signalling pathways would therefore suggest extensive cross-talk between pathway modules and transcriptional networks with a high degree of feedback and feed-forward control taking place. It would also predict that many proteins/genes contribute to the networks functional activity and robustness. This is born out by the fact that pathogens are known target many different molecules in their effort to evade or subvert the immune response [103
], and polymorphisms in numerous proteins has also been shown to be implicated in susceptibility to infectious and inflammatory disease [105
]. Furthermore, experimental modulation of these systems suggests that their activity is directly regulated by numerous factors acting at different levels [109
]. Therefore how real or useful the concept of a bow-tie structure is for such pathways especially with respect understanding their robustness is highly questionable. Indeed the network of molecular interactions that make up the immune system is clearly highly susceptible to perturbation by numerous factors acting at different levels within the network architecture. It has evolved through constant challenges by a diverse range of pathogens. Its robustness relies on the fact that on a population scale its response in different individuals is varied such that what in one may kill, in another may be tolerated.
The task of assembling this diagram has been time consuming and laborious involving 1,000's of hours of work. On the other hand, it summarizes the results of investigations that have taken many times that amount of time to perform and it is difficult to envisage how one could précis this body of work in any other meaningful way. To gain a systems level view of these pathways is to gain an insight into the molecular networks that regulate normal immune function and whose malfunction or manipulation underpins inflammatory and infectious disease. Greater understanding of the overall architecture of the immune system and its susceptibility to deregulation by pathogens and other disease causing agents, should ultimately lead to new strategies and targets for therapeutic intervention. Apart from summarizing decades of research, pathways depicted with formalized graphical notation schemes should aid the communication and comparison of biological data. During a thorough process of internal critiquing sections of the pathway were presented to others who were familiar with the notation scheme but not involved in constructing the pathway presented to them and asked to interpret the biology shown. This process ensured that the interactions of the pathway were not ambiguous in their depiction. Another major incentive for generating pathways with standard notations is to permit the conversion of graphical models into computationally tractable ones, suitable for analysis and simulation. For this purpose we have been exploring the use of signalling Petri nets (SPN) [89
] for modelling "flow" in the integrated pathway diagram. The approach is suited to large scale models and the mEPN scheme used to construct the pathway is easily adaptable into a SPN.
For us the exercise of pathway construction has provided a resource for training, pathway modelling, literature/data interpretation, hypothesis generation and as such is now central to our ongoing investigations of macrophage biology. Importantly however, the pathway model presented here also serves as a worked example of how pathways might be represented in a logical, unambiguous and biologist-friendly fashion, whatever the system of interest. What we would like to see and believe is essential, is the support of the wider community in assembling and editing such diagrams. Such efforts are already underway [22
] and are already providing a vital forum for debate on the known details of pathways in different cell systems. Ideally these efforts will result in detailed models of biological systems that can be shared and assimilated. However, in order to achieve this end pathway models clearly need to be assembled using standard rules and graphical languages. We therefore hope this work will contribute to the ongoing community effort to develop such standards [29