Ingestion-controlling network: what's language got to do with it?

Abstract

The prevalence of obesity has increased dramatically worldwide, whereas the types of treatment and their efficacy have not substantially changed over the last two decades. Additionally, drugs used to control weight gain could occasionally create untoward effects in cardiovascular functions, as well as in behaviors, memory, sleep, and emotions because the molecular machinery responsible for ingestion control is interconnected with or shared by the above domains. How each group of drugs preserves the privacy of its message in the mutual network is not fully understood. In the present essay, the graph theory approach was used to explore some aspects of molecular signaling as though they were a 'language'. Its emphasis is on 'molecular polysemy', a term that refers to the ability of biomolecules to be used like words in natural languages more than one-way. This has physiological and clinical implications, in particular when planning drug designs with "specially engineered shotgun loads" that target a combination of biomolecules that assure a better therapeutic outcome without causing deficits in connected but patho-physiologically irrelevant bystanders.

Fig. 1

A bar chart of functions generated with IPA (Ingenuity R Systems, www.ingenuity.com) functional analysis that implicate ingestions-control peptides. Taller bars are more significant than shorter bars. The probability that each biological function assigned to the set of molecules is due to chance alone was determined using Fischer’s exact test. Y-axis represents the significance score (negative log of p value; B-H: Benjamini-Hochberg method was used for multiple testing corrections). The vertical line denotes the cutoff for significance (p-value of 0.05).

Fig. 2

Molecules isolated during functional analyses as playing statistically significant roles in functions, such as ‘Cellular Growth and Proliferation’, ‘Feeding’ and ‘Behavior’ (see legend on the right). Here, as well as in Fig. 1, functions that exceed the random expectation of their probability are shown in p-values based on Fisher’s exact test. A sample of functions shown in IPA overlay indicates the same molecules a number of identical biomolecules recruited for Vascular development, Memory, Immune response, and Affiliative behavior. These roles were arbitrarily selected from a wider display of available in the overlay macro in order to emphasize the fact they evolved at dissimilar time in individual development.

Fig. 3

A plot testing Zipf distribution using Hermetic Word Frequency Counter¹ based on ranking 850 terms of 91,322 scanned items from IPA-generated studies pooled together. Points are the experimental data and straight line with a negative slope is the best fit to Zipf’s law. Note a group of terms (shown by arrows) with the same frequency and different rank. Despite an adequate agreement with Zipf’s law first five top-ranking datapoints, (LEP, NPY, CRH, POMC, and INC) pull the left tail of the curve down. Abbreviations: CRH - corticotropin releasing hormone; INS – insulin; LEP – leptin; NPY – neuropeptide Y; POMC – proopiomelanocortin. ¹(http://www.hermetic.ch/wfca/wfca.htm)

Fig. 4

Venn diagram with three sets that represent in the various distinct areas of overlapping circles all possible combinations of the relevant molecules (in %) playing statistically significant roles in corresponding functions (‘Behavior’, ‘Immune response’, and ‘Cellular Growth and Proliferation’).

Fig. 5

(A) Biomolecules playing statistically significant roles in ‘Cellular Growth and Proliferation’, ‘Feeding’ and ‘Behavior’; (see Fig. 2 and legend) connected in a single graph (in autolayout). All duplicates molecules (Venn’s “overlap”) were eliminated when submitted to ‘connect’ macro. (B) Maximal triad CRH, LEP, and NPY oversized for visibility along with AVP and POMC in a toggle subcellular layout. Note that in distinction from the classical Venn computations, network rendering emphasized the role of interconnections that includes all relevant information such as the layout constraints for specific functions and type of reactions involved. Arrowheads signify direct interactions between nodes. Lines with a terminal bar indicate negative regulation. In order to reduce the clutter, self-interaction arcs (loops) were removed. Edge labels: A- Activation; B – Binding; - expression (includes metabolism); I - inhibition; LO - localization; P – Phosphory-lation/Dephosphorylation; PP - protein-protein binding; T – transcription