Figure 1. (a) Microfluidic, array-based, and nanomaterials platforms are designed in our laboratory. These assays have been used to study (b) multivalent ligand-receptor binding on supported bilayers, (c) displacement of interfacial proteins involved in the clotting cascade, and (d) the influence of osmolyte and salts on protein denaturation.

Platform Design

The Cremer group has developed a wide variety of high throughput, low sample volume techniques for highly selective analyte sensing. These assays were designed to abstract molecular level details and mechanistic insights from biophysical and bioanalytical problems that would otherwise be difficult to approach. The basic strategy behind integrating on-chip and nanomaterial platforms with traditional physical methods is laid out in Ann. Rev. Phys. Chem., 56, 2005, 369-387. For example, we invented temperature gradient microfluidics (JACS, 124, 2002, 4432-4435), optical methods for interfacial patterning of biomolecules from aqueous solutions under mild conditions (JACS, 125, 2003, 8074-8075), and air-stable/size-selective supported phospholipid membranes (JACS, 126, 2004, 6512-6513; JACS, 128, 2006, 7168-7169; JACS, 129, 2007, 10567-10574). We also developed a method for spatially addressing phospholipid bilayer arrays on planar supports (Nature, 413, 2001, 226-230; JACS, 121, 1999, 8130-8131).
One of our most recent advances has been to develop the use of supported bilayers as a novel matrix for the separation of membrane components within a bilayer (JACS, 129, 2007, 10567-10574). Finally, we have designed rapid prototyping methods for the production of highly uniform patterned lines and nanodots on planar substrates at the 10 to 100 nm scale (Adv. Mater.,18, 2006, 2240-2243; Nano Letters, 7, 2007, 2452-2458). The advantage of our nanopatterning techniques is that they can be performed under ambient conditions within a few minutes and without the need for complex instrumentation.

Multivalent Ligand/Receptor Binding

One major area of focus in the Cremer laboratory has been in understanding multivalent ligand-receptor binding between phospholipid membranes and aqueous proteins containing multiple binding sites (JACS, 125, 2003, 4779-4784). Phospholipid membranes are two-dimensionally fluid; therefore, membrane-bound ligands can undergo constant rearrangement within the bilayer. Such rearrangements are especially important for binding proteins from aqueous solution. We performed microfluidic studies which clearly demonstrated that bivalent antibody molecules became increasingly tightly bound at the interface as the ligand density in the lipid membrane was increased. By contrast, we were also able to demonstrate that the binding between the B subunits of cholera toxin (CTB) and GM1, a membrane glycolipid, became substantially weaker as the GM1 concentration in the bilayer was increased (JACS, 129, 2007, 5954-5961). Further studies using atomic force microscopy showed a direct correlation between the weakening of the thermodynamic binding constant and the clustering of GM1 molecules (Figure 2). Indeed, hydrogen bonding between adjacent glycolipids almost certainly hindered the binding of CTB. Such a result is significant because GM1 has been shown to preferentially partition into lipid raft regimes in model membrane systems. Clustering may be a useful strategy to enhance the binding of proteins from aqueous solution in some cases. Our results, however, clearly indicated that this need not always be the case.

Figure 2. Schematic representation of the inhibition of CTB binding by GM1 clustering on supported POPC bilayers. For simplicity GM1 molecules in the lower leaflet are not drawn.

We have also used our microfluidic assays in collaboration with biologists to elucidate the binding of chemoattractants to E. coli (PNAS 100, 2003, 5449-5454) as well as the multivalent binding of three-domain Bacillus thuringiensis (BT) crystal (Cry) protein toxins to complex glycolipids (Science 307, 2005, 922-925). In the latter case, we worked with a team of investigators from the United States and Britain to demonstrate the specificity of BT toxins for glycolipid molecules presented on the surface of intestinal cell membranes of C. elegans. In particular, it was found that proteins such as Cry5B bound specifically with glycolipid molecules containing a core tetrasaccharide motif, which is specific to invertebrates. This suggests that it may be possible to employ Cry toxins to control parasitic nematodes in animals and plants used in the agriculture industry. Moreover, these proteins have no toxicity in humans or other vertebrates that lack the conserved tetrasaccharide core. A key aspect of this research involved determining the equilibrium dissociation constants for Cry toxins to glycolipids presented in supported lipid bilayers coated on the inside of microfluidic devices. The glycolipids, which needed to be reconstituted from C. elegans, were only available in trace quantities (typically in the picomole range). It would have been virtually impossible to obtain thermodynamic data from these systems without the development of our on-chip platforms.

Macromolecule Adsorption and the Vroman Effect

We have investigated the adsorption of protein, surfactant, polymer, and lipid membrane structures at solid/liquid interfaces (JACS, 128, 2006, 5516-5522; JACS, 125, 2003, 11166-11167; JACS, 125, 2003, 12782-12786; JACS, 124, 2002, 8751-8756; Science, 293, 2001, 1292-1295; JACS, 122, 2000, 12371-12372). An excellent example of this research concerns the use of atomic force microscopy, nonlinear optical techniques, and microfluidic platforms to investigate the interactions between the blood protein, fibrinogen, and oxide surfaces. Fibrinogen conversion to fibrin is the penultimate step in the clotting cascade. It is therefore crucial to understand how this protein interacts with artificial materials placed in the body (e.g. hip implants). It was noted in the late 1960s by Leo Vroman and coworkers that fibrinogen is often the first biomolecule to adsorb at oxide and polymer surfaces. It can, however, be displaced under certain circumstances by other more benign proteins in blood. The mechanism by which this occurred had remained unknown for three and one-half decades until we showed that fibrinogen's positively charged ?C domains are the key to the process. Specifically, weak salt bridges are formed between these Lys and Arg rich domains and an underlying oxide surface. Such interactions prevent the rest of the protein from making more substantial contacts (e.g. hydrogen bonding, van der Waals, and hydrophobic interactions) with the underlying substrate (Figure 3).

Figure 3. Left: A fibrinogen molecule adsorbed to an oxide surface. Its ?C domains (in light green) sit directly on the substrate surface beneath the molecule’s central E domain (in red). Under this configuration it is readily displaced from the interface. Right: The ?C domains have migrated to a position above the central E domain. In this case the biomacromolecules binds tenaciously to the interface.

Hofmeister and Osmolyte Studies

We have investigated the effect of salts and osmolytes on the folding of proteins and colloidal structures in aqueous solution (JACS, 2007, in press; JACS, 127, 2005, 14505-14510; JACS, 126, 2004, 10522-10523; JACS, 125, 2003, 15630-15635; JACS, 125, 2003, 2850-2851). It was noted as early as 1888 that the identity and concentration of the anions present in solution profoundly influences a wide variety of macromolecular phase transitions. Indeed, the ability of a given anion to affect everything from protein crystallization to nanomaterial assembly in aqueous solution generally follows a recurring trend known as the Hofmeister series:

CO32- > SO42- > S2O32- > H2PO4- > F- > Cl- > Br- ~ NO3-> I- > ClO4- > SCN-

Figure 4. Interactions amongst anions, PNIPAM, and hydration waters: (a) Hydrogen bonding of the amide and its destabilization through the polarization of water by the anion, X-. (b) The hydrophobic hydration of the molecule is associated with surface tension. The surface tension can be modulated by adding salt. (c) Direct binding of the anion to the amide group of PNIPAM.

In closely connected studies, we have explored the mechanism by which urea and related osmolytes can denature or stabilize folded proteins (JACS, 2007, submitted). In contrast to Hofmeister salts, osmolytes are often present in solution at much higher concentrations. For example, 8 M urea solutions are commonly used for denaturation. There has been a long standing question as to whether the mechanism by which urea denatures biomacromolecules is direct or indirect. A direct mechanism would proceed via hydrogen bonding between the osmolyte and the protein, while an indirect mechanism could putatively involve the disruption of interfacial water structure. Again, a combination of microfluidic platforms and nonlinear optical techniques have been employed in our laboratory to investigate this problem. Our results allow us to discount the direct mechanism. We are presently exploring the pathway by which disrupted water structure could lead to denaturation.