Hydrophobin Bilayer as Water Impermeable Protein Membrane

Abstract

One of the most important properties of membranes is their permeability to water and other small molecules. A targeted change in permeability allows the passage of molecules to be controlled. Vesicles made of membranes with low water permeability are preferable for drug delivery, for example, because they are more stable and maintain the drug concentration inside. This study reports on the very low water permeability of pure protein membranes composed of a bilayer of the amphiphilic protein hydrophobin HFBI. Using a droplet interface bilayer setup, we demonstrate that HFBI bilayers are essentially impermeable to water. HFBI bilayers withstand far larger osmotic pressures than lipid membranes. Only by disturbing the packing of the proteins in the HFBI bilayer is a measurable water permeability induced. To investigate possible molecular mechanisms causing the near-zero permeability, we used all-atom molecular dynamics simulations of various HFBI bilayer models. The simulations suggest that the experimental HFBI bilayer permeability is compatible neither with a lateral honeycomb structure, as found for HFBI monolayers, nor with a residual oil layer within the bilayer or with a disordered lateral packing similar to the packing in lipid bilayers. These results suggest that the low permeabilities of HFBI and lipid bilayers rely on different mechanisms. With their extremely low but adaptable permeability and high stability, HFBI membranes could be used as an osmotic pressure-insensitive barrier in situations where lipid membranes fail such as desalination membranes.

Figure 1

Droplet interface bilayer (DIB) experiments. (a) Sketch from the side of the experimental setup of two HFBI-coated buffer droplet pairs of different salt concentrations on a PDMS-covered glass substrate. (b, c) Side and top views of two HFBI-coated buffer droplets brought into contact and forming a DIB. The osmotic concentration difference of the two droplets was 1.717 osmol/L. The water permeability was measured by observing the volume changes of the individual droplets over time. The contour of the droplet pairs (orange) at the beginning of the measurements was copied and pasted into the image taken at 6 min. The scale bar indicates a size of 500 μm. (d) Volume change of droplets over time with pairs of monoolein (green circles) and HFBI (blue triangles) coated droplets. In both measurements, the osmotic concentration difference between the two droplets in contact was 0.259 osmol/L. The inset shows an enlargement of the volume change of the HFBI-coated droplet pair for the last 2 min. The red line indicates zero volume change.

Figure 2

(a) Volume change in percent of droplets, with initial size of V₀ = 0.7–0.9 mm³, over time with droplet pairs having a HFBI-dCBM:HFBI WT ratio of 0:1 (blue triangles) and 0.4:1 (purple stars). (b) Box and whisker plots (min-to-max) of the mean permeability values of several HFBI membranes in the presence of the mutant HFBI-dCBM in different weight ratios. Osmotic concentration difference: 1.717 osmol/L for pure HFBI membranes and 0.086 osmol/L for HFBI-dCBM:HFBI mixtures. Temperature: 30 °C.

Figure 3

HFBI monolayer simulations. (a) Side view of simulation system shown in surface representation composed of proteins (green) with hydrophobic patches (orange) and water (blue). The simulation box is shown as a dotted black line. (b) Mass density along the membrane normal averaged over four independent simulations. (c) First lateral box dimension versus time taken from four independent simulations (shaded areas). Lines show running averages to guide the eye.

Figure 4

(a) Top: top view of the simulation setup of HFBI with a honeycomb structure. One simulation cell is colored dark green. Light green regions depict periodic images of the cell. Bottom: schematic view of the “dense” overlay of two laterally displaced monolayers with cavities spanning only one leaflet. (b) Example of expansion of the first lateral box dimension during simulations based on the HFBI-α (red) and HFBI-β (blue) unit cells, composed of either the “dense” or the “holey” membrane model (see legend).

Figure 5

Mass densities of HFBI (green), hydrophobic patches (orange), and water (blue) for the “dense” HFBI bilayer based on the HFBI-β unit cell taken from different time intervals 0–1 ns (left), 10–11 ns (middle), and 100–101 ns (right). Contributions from the upper and lower leaflet are plotted in different shades (see legend).

Figure 6

(a) Scheme of the HFBI structure. Secondary structure is shown in green and the hydrophobic patch in orange including the side chains. In addition, the charged side residues Asp³⁰ and Lys³² and the polar side chain of Gln⁶⁵ are highlighted in red and blue, respectively. (b) Graphical representation of a dense disordered HFBI monolayer (top view) after compression procedure. (c) Snapshots (side view) from two different time points of a HFBI bilayer simulation build of two monolayers as represented in (b). Water is shown in blue.

Figure 7

Distribution of permeability values calculated for every single time step for three pure HFBI wild-type membranes.