Role of lipid-protein interactions in insulin signaling and diabetes
PhD student: Fatma Betül Can Yakac
Supervisor at TUD: Ünal Coskun
Supervisor at KCL: Francesco Rubino
Start date: 15.10.2017
Bariatric surgery of obese patients coincides with an acute increase in lipolysis that presumably leads to the remodelling of serum fatty acid profiles. Surprisingly, some patients show remission of their Type 2 Diabetes within a few days. However, whether the observed serum fatty acid changes post-surgery have an effect on cellular membrane lipidome remodelling and concomitant changes in insulin signalling and downstream signalling were not studied / shown conclusively.
In multicellular organisms, insulin signalling fulfils essential cellular functions far beyond the regulation of glucose metabolism. The binding of its cognate ligands leads to concomitant activation of the insulin receptor tyrosine kinase domain in-trans and subsequent recruitment of several downstream signalling effector proteins, such as the oncogene RAS, phosphoinositide-3 kinase (PI3K), phospholipase C (PLC), phosphatase and tension homologue 10 (PTEN), phosphoinositide dependent kinase (PDK) and AKT. AKT is a central player in insulin signalling as it is downstream of the insulin receptor and upstream to GLUT trafficking and thus activation. In addition, AKT is a negative regulator of the transcription factor FOXO.
Recent data from the Coskun laboratory shows that in Drosophila melanogaster, AKT isoform activity / selective phosphorylation can be regulated in response to dietary lipids and that the lipid binding pleckstrin homology domain (PH-domain) of human AKT is potentially able of discriminating PI(3,4,5)P3 lipid species.
For my PhD thesis I aim at studying: i) whether membrane lipidome changes induced by fatty acid feeding can alter AKT activity and thus insulin signalling outcome in hepatocytes (HepG2 cells); and ii) the molecular determinants of AKT membrane recruitment using synthetic membrane systems and purified proteins.
To assess lipid-binding preferences of AKT1-PH domain, three recombinant constructs were expressed in E.coli BL21 (DE3) strain with N-terminal cleavable GST-tag: 1) AKT1-PH with C-terminal sortase tag (AKT1-PH-sortase), 2) a monomeric EGFP fusion protein of AKT1-PH domain (mEGFP-AKT1-PH), and 3) AKT1-PH domain 8 Histidine residues at the C terminus (AKT1-PH-His8).
Protein purification was based on affinity chromatography using Glutathione Sepharose beads, followed by proteolytic cleavage of tags, and completed by size exclusion chromatography (SEC) on Superdex 75 column to isolate the monomeric population of the respective proteins. Although majority of the proteins was eluted in the void volume during SEC as aggregates, we obtained highly pure, monomeric and soluble ~42 kDa mEGFP tagged AKT1-PH and ~16kDa of AKT1-PH-sortase and AKT1-PH-His8 proteins as observed by Coomassie stained SDS-PAGE. Furthermore, size distribution of the proteins was checked by dynamic light scattering (DLS).
To characterize the interaction of AKT1-PH with PI(3,4,5)P3 containing membranes, first giant unilamellar vesicles (GUVs) were used. They are free-standing membranes with size similar to that of cells and thus one can monitor the interactions by confocal fluorescent microscopy. Additionally, membrane binding of proteins is usually performed under equilibrium conditions, and can be monitored on individual vesicles. This allows us to correlate the actual binding with vesicle size and by applying additional probes, also with membrane order.
GUVs were prepared by polyvinyl alcohol (PVA) assisted method with defined lipid composition of POPC/PI(3,4,5)P3 (95/5 mol%); one of the 3 commercially available synthetic lipid species of PI(3,4,5)P3 was used: dipalmitoyl (16:0/16:0), dioleoyl (18:1/18:1) or stearoyl/arachidonic acid (18:0/20:4)) and contained a lipophilic membrane dye DiI (0.05 mol%). The binding of mEGFP-AKT1-PH protein was monitored by confocal microscopy within 15-30 minutes after protein addition. The binding efficiency was quantified at individual GUVs level based on fluorescent signal of EGFP. According to GUV data, protein was only bound to membranes with PI(3,4,5)P3 species, and no binding was observed on POPC-only containing vesicles. Importantly, mEGFP-AKT1-PH domain showed increased mean signal intensity for unsaturated and longer fatty acyl chains of stearoyl/arachidonic (18:0/20:4) and dioleoyl species (18:1/18:1) in comparison to saturated dipalmitoyl (16:0/16:0) species of PI(3,4,5)P3.
Liposomes with defined size of 100 nm, composed of 95% POPC and 5% PI(3,4,5)P3 (one of the 3 commercially available synthetic lipid species of dipalmitoyl (16:0/16:0), dioleoyl (18:1/18:1) or stearoyl/arachidonic acid (18:0/20:4)) were used. The liposomes were prepared and characterized for composition, charge and radius using thin layer chromatography (TLC), zeta potential measurement and dynamic light scattering respectively.
I aimed at establishing a fast and direct binding assay by microscale thermophoresis assay, which measures directed movement of fluorescent molecules through microscopic temperature gradient. The signal of fluorescent reporter changes upon interaction, and can then be used to quantify the binding equilibrium constant of interacting molecules.
For the assay optimization, binding of AKT1-PH-His8 protein, labelled with tris-NTA-Alexa647 dye as fluorescence reporter, to liposomes containing 18:0/20:4 PI(3,4,5)P3 was used. The determined equilibrium binding constant (KD) showed that PH domain of AKT1 selectively binds to its lipid target (18:0/20:4) PI(3,4,5)P3 with KD around 1.5 M, whereas no saturation curve was obtained with POPC itself. These highly promising result now enable me to measure and comprehensively compare binding properties of AKT1-PH domain to different species of PI(3,4,5)P3 containing liposomes with orthogonal lipid-protein binding assays.
Finally, I plan to biochemiacal characterize the AKT1-PH domain, determine the lipid-binding preferences of AKT1-PH domain and the effect of other lipids such as cholesterol on protein binding preferences to lipid species of PI(3,4,5)P3.