Cell Bi­ol­ogy

The aim of the project is to ex­plore and re­con­struct the bi­o­log­i­cal NF-κB tran­scrip­tion fac­tor sig­nal trans­duc­tion sys­tem for bet­ter un­der­stand­ing. In par­tic­u­lar we plan to map out the mech­a­nisms that con­vert a chem­i­cal stim­u­lus to a cell into a spe­cific cel­lu­lar re­sponse in the ex­am­ple of neural stem cells and adult neu­rons in mice. 

Fig. 12: Analy­sis on the in­te­grated NF-κB protein-​protein in­ter­ac­tion and sig­nal­ing trans­duc­tion net­works with edge bundling tech­niques and graph the­ory. © Uni­ver­sität Biele­feld

Due to the com­plex­ity of path­way in­ter­ac­tions and large num­bers of com­po­nents in­volved in cell pro­lif­er­a­tion, cell dif­fer­en­ti­a­tion, sig­nal trans­duc­tion, cel­lu­lar rhythms, and cell-​to-cell com­mu­ni­ca­tion, it is quite dif­fi­cult to in­tu­itively un­der­stand the be­hav­ior of cel­lu­lar net­works. Re­cent ex­per­i­men­tal and com­pu­ta­tional progress yields net­works of in­creased size and com­plex­ity that need to be ex­am­ined care­fully. 

A com­mon way to ac­cess the in­for­ma­tion in a net­work is through di­rect vi­su­al­iza­tion, but this fails as it often only re­sults in “fur balls” from which lit­tle in­sight can be gath­ered. There­fore we dis­cover a new vi­su­al­iza­tion ap­proach, to high­light and com­mu­ni­cate one par­tic­u­lar piece of in­for­ma­tion about dy­namic net­work struc­tures. More­over, we en­deavor to find a loss­less trans­for­ma­tion of dy­namic sig­nal­ing net­works into a com­pact, less re­dun­dant rep­re­sen­ta­tion. We are in­ves­ti­gat­ing novel rep­re­sen­ta­tions of net­works, which re­duce net­work com­plex­ity by ex­plic­itly rep­re­sent­ing re-​occurring net­work mo­tifs and dy­nam­ics, with­out any loss of in­for­ma­tion. 

As a first re­sult, we present a new ap­proach to net­work vi­su­al­iza­tion that tightly in­te­grates net­work analy­sis meth­ods and edge bundling tech­niques. The ap­proach uses edge cen­tral­ity mea­sures to drive a force di­rected edge bundling method; this re­sults in pic­tures that clearly show the most sig­nif­i­cant topo­log­i­cal skele­ton struc­tures of the input net­work. We also in­tro­duce a new force-​directed ra­dial lay­out that shows group analy­sis of the k-​cores: this re­sults in pic­tures that show the im­por­tant co­he­sive sub­group struc­tures of the input net­work and their re­la­tion­ships (see Fig. 12). 

The foun­da­tional mod­el­ing ideas and the re­la­tion­ship be­tween net­work struc­ture and sys­tem dy­nam­ics is a rapidly ex­pand­ing fron­tier of this new sci­ence. Of par­tic­u­lar in­ter­est is the un­der­stand­ing of the or­ga­ni­za­tion, com­plex­ity and dy­nam­ics of bi­o­log­i­cal net­works and how these are in­flu­enced by net­work evo­lu­tion and func­tion­al­ity. 

We aim at de­vel­op­ing a the­o­ret­i­cal frame­work for the func­tion of spe­cific sig­nal in­te­gra­tion mod­ules in the local and global net­work con­text. The­o­ries are in de­vel­op­ing to ad­dress spe­cific as­pects of the net­work func­tion, rang­ing from, for ex­am­ple, (1) graph­i­cal cell-​based mod­els to de­scribe de­vel­op­ment and (2) net­work con­struc­tion based on data and text min­ing to (3) quan­ti­ta­tive mod­els with im­ple­mented re­ac­tion ki­net­ics and mass ac­tion re­la­tion­ships to de­scribe in­for­ma­tion flow. 

In order to re­con­struct and an­a­lyze the NF-κB sys­tem we have begun to make use of data min­ing and in­for­ma­tion fu­sion by in­te­grat­ing data­bases with dif­fer­ent con­tents: e.g. in­ter­act­ing part­ners and tar­get genes. For the most im­por­tant el­e­ments within the sys­tem we have gen­er­ated ex­per­i­men­tal and data­base gen­er­ated net­work sys­tems that are reg­u­lated by the NF-κB tran­scrip­tion fac­tor. The col­lected data and in­for­ma­tion is a com­bi­na­tion of tightly in­ter­linked com­plex sys­tems at var­i­ous lev­els of mag­ni­tudes. 

Each net­work is con­structed using a few basic mech­a­nis­tic mo­tifs/mod­ules. The func­tion­al­ity of each net­work im­pli­cates par­tic­i­pa­tion of spe­cific in­ter­act­ing pro­teins, where some pro­teins ex­e­cute few, other many in­ter­ac­tions. The net­works op­er­ate in the di­men­sion of time and net­work per­for­mance in agree­ment with cell re­quire­ments de­mands reg­u­la­tion of its ac­tiv­ity, e.g. by feed­back mech­a­nisms to en­able re­li­able cell fate de­ci­sions. Net­work reg­u­la­tion de­pends on ad­di­tional in­ter­ac­tions that are clearly vis­i­ble in so called mul­ti­di­men­sional net­works. 

The con­structed mod­els are ex­am­ples of how the be­hav­ior of cells mod­eled by a Petri net can be sim­u­lated (see Fig­ure 12

Fig. 13: Sim­u­la­tion of the NF-κB sys­tem in the Petri Net lan­guage. © Uni­ver­sität Biele­feld

Ideas for lab­o­ra­tory ex­per­i­ments can be gained and tested by chang­ing the basic Petri net. One ap­proach is a the­o­ret­i­cal knock­out ex­per­i­ment. For this task, knocked out genes are mod­eled by delet­ing the cor­re­spond­ing tran­si­tions. A change of gene se­quences in­flu­enc­ing the cat­alytic ef­fi­ciency can be mod­eled by chang­ing speed pa­ra­me­ters of the cor­re­spond­ing tran­si­tion. Fac­tors out­side the cells that in­flu­ence the sig­nal mol­e­cules, the dif­fu­sion speed of the medium, and the avail­abil­ity of nu­tri­ents or even processes in­side the cell can also be mod­eled.