A basis for a visual language for describing, archiving and analyzing functional models of complex biological systems

Abstract

Background: We propose that a computerized, internet-based graphical description language for systems biology will be essential for describing, archiving and analyzing complex problems of biological function in health and disease.

Results: We outline here a conceptual basis for designing such a language and describe BioD, a prototype language that we have used to explore the utility and feasibility of this approach to functional biology. Using example models, we demonstrate that a rather limited lexicon of icons and arrows suffices to describe complex cell-biological systems as discrete models that can be posted and linked on the internet.

Conclusions: Given available computer and internet technology, BioD may be implemented as an extensible, multidisciplinary language that can be used to archive functional systems knowledge and be extended to support both qualitative and quantitative functional analysis.

Figure 1

A basic lexicon of icons and arrows for describing the function of complex biological systems using BioD. Object icons are provided for common biological structures such as atoms, molecules, molecular sites and compartments, and for events and processes. 'Action' arrows represent the actions that the functional properties of one object can have on the functional properties of other objects and actions.

Figure 2

(a) A simplified model of a Cdk4 kinase molecule illustrates how basic BioD icons and action arrows can concisely represent intra- and inter-molecular actions. The Cdk4 molecule includes a kinase site (K) that, when active, phosphorylates a phosphorylation site (P) on the RB protein. The kinase site on Cdk4 is activated (filled arrow) by occupancy of its phosphorylation site and inhibited (open-squared arrow) by occupancy of the binding site (dotted circle) that binds the Cdk4 inhibitor p15. (b) An 'event model' derived from the model above. Events are defined as changes of state of one or more functional properties of icons in a state model. Here, for instance, the event model displays a chain of events triggered by an increase of p15 concentration (see text).

Figure 3

A BioD model of structures and functional elements controlling the G1/S transition of the cell cycle [6,10,11,12,13,14]. Key molecular players are represented by binding sites, phosphorylation sites and kinase sites whose occupancies and activities interact to control the cell cycle. The critical synthesis and degradation kinetics of cyclin D1 are represented by an 'identity' link to the nuclear cyclin D1 icon in the 'process' modeled in Figure 4. Additional inputs to the model (TGF-β stimulation and DNA damage) and outputs (G1/S transition) are included as unmodeled processes. CAK, cyclin-activating kinase; DP, DRTF1-polypeptide; DRTF, differentiation-regulated transcription factor; E2F, E2F transcription factor; TK, thymidine kinase; TS,thymidylate synthetase; POL, DNA polymerase.

Figure 4

The activation of cyclin D1 gene expression by parallel Ras kinase cascades [44] modeled with BioD. Nuclear cyclin D1 levels depend on cytoplasmic-nuclear transport [45] and cytoplasmic degradation of cyclin D1 via an unmodeled ubiquitin proteolysis ('Ub proteolysis') process. The cyclin D1 icon in this model is identified with the cyclin D1 icon in Figure 3 using an identity link (Fig 3.cyclin D1; which is reciprocal to the Fig 4.nucleus.cyclin D1 link in Figure 3). The inhibition of nuclear export of cyclin D1 is inhibited by an 'action' link (Fig 5.CaM.action) that allows activated calmodulin (CaM) in the model of Figure 5 to reach into and affect this model. ERK, extracellular-regulated kinase; G3K3β, glycogen-3-synthase kinase-3β; MEK, MAP/ERK kinase; PI3K, phosphatidylinositol trisphosphate kinase.

Figure 5

A model of the activation of calmodulin (CaM) by calcium influx in which K-channel blockers [46,47,48] inhibit ATP-sensitive potassium channels (KATP) and thus cause membrane depolarization (depol). Membrane depolarization activates voltage-dependent calcium 'ion channel' transporters and Ca²⁺ influx which have the dual effects of increasing intracellular Ca²⁺ concentration and further increasing depolarization. The action of Ca²⁺-activated CaM is linked via an action link (Fig 4.nucleus.cyclinD1.export) to the inhibition of cyclin D1 export in Figure 4.

Figure 6

The use of 'inhibit' and 'activate' arrows to represent steric hindrance and cooperative binding reactions in a model of the lysis-lysogeny decision network of phage lambda [15]. A single-molecule icon is used to represent transcription factors that are actually homodimers of CI, Cro and CII. Key pathways for the synthesis and degradation of CI and Cro are as follows. Binding of RNA polymerase (RNAP) to the PRM promoter activates production of CI via the cI gene. The binding of 'RNAP' (the quotation marks signify that the icon is a duplicate of the RNAP icon and is not a separate entity) to promoter PR activates production of cro RNA via the cro gene. Binding of CII to promoter PRE at the opposite end of the cro gene activates production of antisense cro RNA ('anti-cro'). cro RNA required to produce Cro protein is degraded and removed from the system by a quenching reaction with anti-cro. The degradation of CI (the vertical reaction arrow at the top of the CI icon) depends on the proteolytic enzyme RecA, which is activated by ultraviolet light (UV). Cro degradation is unregulated, and CII degradation is not represented. The diagram is laid out to emphasize the symmetry between the CI and Cro synthesis and degradation pathways. The symmetry is broken, however, when one considers how competitive binding of the transcription factors (CI and Cro) to the operators (OR1, OR2 and OR3) controls access of RNA polymerase to the back-to-back promoters PRM and PR. Transcription repression by CI and Cro is nearly symmetric as they each limit access (by steric hindrance; shown as double-headed 'inhibit' arrows) of RNA polymerase to the promoters. RNA polymerase access to PR is limited in four instances by CI or Cro binding to either OR1 or OR2, whereas access to PRM is limited in three instances by Cro binding to either OR2 or OR3, or by CI binding to OR3. The break in symmetry occurs because, rather than being repressive, CI binding to OR2 actually enhances PRM transcription, which is stabilized in two ways. First, CI bound to OR3 directly binds RNA polymerase to stabilize its binding at its promoter site. Second, CI stabilizes its own binding to OR2 by establishing a homodimerization bond (the dimerization arrow extending to the right of the CI icon) with a CI molecule bound to OR1. The result is two trimers (CI-gene-RNAP and CI-gene-CI) within which the ability of pairs of bonds to stabilize a third bond is indicated by a double-tailed activation arrow. As such bond stabilization is mutually cooperative, the three double-tailed arrows are superimposed to form the triadic arrows shown in the figure.