Surface modification of an ultrafiltration membrane with crystalline structure and studies on interactions with selected protein molecules

Abstract

The crystalline cell surface layer (S-layer) from the Gram-positive bacterium Bacillus sphaericus CCM 2120 which shows square symmetry and a pore size of 4–5 nm was used for production of ultrafiltration (UF) membranes, termed S-layer ultrafiltration membranes (SUMs). Since S-layers are composed of identical protein or glycoprotein subunits which are arranged according to a 2D protein crystal, pores passing through show identical size and morphology. SUMs were produced by depositing isolated S-layer material on microfiltration membranes, crosslinking the S-layer protein with glutaraldehyde and reducing Schiff bases with sodium borohydride. In native S-layer lattices equimolar amounts of amino and carboxyl groups were found to be exposed on surface-located protein domains and in the pore areas. After crosslinking the S-layer protein with glutaraldehyde, the S-layer surface and the pores assumed a net negative charge. By using S-layer material in which only the outer face of the S-layer lattice was exposed 1.6 carboxyl groups per nm 2 were determined which are relevant for interactions with protein molecules in solutions. For preparing ultrafiltration membranes with crystalline structure and high charge density, glutamic acid residues were stepwise introduced into the S-layer lattice which led to a charge density of either 3.2 or 7.0 carboxyl groups per nm 2. Conversion of negatively charged carboxyl groups into free amino groups by preserving the crystalline lattice structure was also possible. For obtaining more detailed information on factors inducing flux losses of UF membranes, neutral hydrophilic and neutral hydrophobic, negatively and positively charged SUMs were used for filtration of selected test proteins differing in molecular size and charge. The combination of filtration experiments and high-resolution electron microscopical studies allowed to determine that adsorption of proteins inside the pores with a size similar to the dimension of the test molecules caused considerable flux decline of up to 70%. As derived from the thickness of the S-layer lattice and the molecular dimension of the test proteins, a single layer of protein molecules inside the pores was sufficient for generating measured flux losses. On the contrary, ferritin which is significantly larger than the pores could only adsorb on the outer face of the S-layer lattice but not penetrate the pore openings. Although electron microscopical studies revealed that a monolayer was present on the S-layer surface, flux losses of SUMs were only in the range of 20%.