The immune response is a complex system with many possible inputs, interactions and outcomes. Structured analysis of the dataflow in these cells holds the promise of both better understanding this system and predicting the response of these cells to a given stimulus. Host/fungal interactions have so far been studied only for pathogenic fungi such as C. albicans
, but it is not known what keeps commensal fungi far from being recognized as pathogenic. To elucidate the mechanism by which the host immune system discriminates friends from foes, we used the specifically tailored data model, the BCML 
and the DC-ATLAS type of community curation 
to annotate the modules that make the transcriptional network for the C-type lectins DC-SIGN and Dectin-1. DC-ATLAS pathways were then used to generate pathway signatures and to compare different microarray experiments in order to obtain a detailed outlook of the genomic response to different fungi or fungal forms over time. Our results show how the accurate annotation of existing knowledge on signaling by Dectin-1 and DC-SIGN can lead to the understanding of the DC response to similar microorganisms with a very different immunological outcome. We were not able to obtain the same results using traditional pathways in public databases. One reason is that core modules of the immune network crucial in this process, such as DC-SIGN and Dectin-1, simply are not available. Secondly, the other pathways (the public counterparts of the ones stored in DC-ATLAS) are not specific for a cell type, and as such contain connections that are absent in DCs, or miss components that actually play a role in the transduction mechanism. Finally, we were able to dissect the flow of information thanks to the structure of the DC-ATLAS, which describes pathways making the immune network as assemblies of three modules: sensing, signal transduction and outcome. This modular nature allows us to further distinguish responses into receptor-specific signaling pathways and with respect to the timing of the signal flow. Thus, DC-ATLAS pathways allowed us to differentiate the temporal and modular events characterizing the signaling cascades from initial fungal recognition to DC maturation. The results of the analysis are thus much improved with respect to the results one can obtain from existing pathway databases, which cannot significantly sort out changes in specific processes because they are too large and redundant. In general, our results demonstrate significant added value when re-analysing already available datasets with this newly designed tool.
Our pathway analyses indicated that the different immune response to Sc yeast cells and Candida hyphae stems from an important difference in the impact on transcription of C-type lectin pathway components. This evidence supports the observation that the activation of the Dectin-1 pathway is probably strongly delayed after stimulation with Candida hyphae possibly due to a slower ligand release and resistance to degradation in the endosome. Indeed Candida hyphae do not readily expose β-glucans on their surface but these may be released slowly upon phagocytic processing leading to low but sustained Dectin-1 signaling. On the other hand, S. cerevisiae cells will signal strongly through Dectin-1 from the moment of initial recognition. Not only the amount of ligand available but also the timing of its release and the location (e.g. plasma membrane or phagosome) may determine the outcome. Our data suggest that the regulation of receptor signaling both in place and over time may balance tolerogenic and inflammatory responses.
Efficient recognition of the stimulus results in an active down-tuning of the receptor as exemplified by the shut-down of Dectin-1 (CLEC7A
gene) expression. This is a new finding, whose statistical assessment is totally dependent on use of the DC-ATLAS logic. The down-regulation of the C-lectin receptor modules as early as 4 hours after stimulation suggests that, at that time point, the cell is already dedicated to priming its response towards fungi. This indicates that the receptor mRNA levels are important for cell homeostasis and that the protein levels are tightly and rapidly controlled by efficient turnover mechanisms. The significant down-regulation, we reported, is in agreement with previous observations that the expression of C-type lectins is highly regulated by maturation stimuli, leading to their down-regulation as DCs mature 
Through the use of DC-ATLAS we were able to discern the differential involvement of receptors that relate to differences in the disposition of cell wall components (i.e, C. albicans hyphae vs Sc yeast cell form), as well as the possibility of distinguishing pathogenic from non-pathogenic fungi. Also, we were able to extract the contribution of single cell components during the response to a whole fungus.
We demonstrated that the integration of multiple stimuli over a defined temporal window is required to obtain an effective response.
In line with the known complex organisation of the fungal cell wall, our analysis reveals how fungal recognition is not simply the product of a series of linear signaling pathways but it is comprised of a complex set of integrated responses arising from a dynamic network of thousands of molecules that are subject to regulatory mechanisms themselves.
We highlight the ability of phagocytosed S. cerevisiae to induce a broader and more intense innate immune response than when different glycans are engaged to their corresponding PRRs alone. Overall, these core responses suggest that it is really the integration of signals, from both TLR and C-type lectins that determines the response to yeast cells rather than a response dominated by one or only few receptors.
Altogether our results show that the signal travels rapidly through the network and affected DCs are committed promptly after signal initiation when DCs rapidly down-regulate their receptor sensing modules and elicit signals stimulating downstream immune effectors depending on ligand concentration and/or availability. Differential activation of DCs by the various fungal stimuli results in different cytokine expression profiles that are related to their pathway profiles. Thus, depending on the type and state of the fungus, DCs respond with different cytokine signals to instruct an adaptive T cell response tailored to that specific fungal threat.
This work is based on in vitro experiments. In vivo the situation is indeed more complex, with the interplay and cross-talk among different immune cells and stromal cells that together determine the final response to the pathogen. Nonetheless we demonstrate that differences between pathogenic or harmless fungal strains exist when it comes to detection by DCs, important players in the initiation and control of the immune response. These findings should be expanded to study of the response of other immune or non-immune cells that also come into contact with these pathogens and that possess similar or other types of pattern recognition receptors.
This work establishes a broadly applicable approach, instrumental in revealing transcriptional and regulatory networks of immune cells upon pathogen recognition, leading to a better understanding of fungal infection in general and in relation to host genetic susceptibility. A better understanding of the modular nature of DCs response to fungi will only come from comparative studies of the immune response to whole cells of pathogenic and non-pathogenic species and strains. Only then we will fully understand how the host immune system detects pathogens.