Structure and Dynamics of Polymeric and Biomolecular Materials at Interfaces

The mission of IRG-2 is to pioneer the fabrication and in situ observation of dynamic interfaces.  
Although inspiration often comes from biomolecular systems such as phospholipid membranes or DNA assemblies, the applications are broad and include lubrication, emulsion stability, and polymer processing. We wish to build on the strong materials and interface engineering capabilities within CPIMA to create a new generation of biological tools and systems that will transform biological experiments on membranes and provide the stepping-stones toward development of advanced materials with new levels of complexity and communication. 

Participants
Boxer, Frank, Fuller, Groves, Longo (Coordinator), Muller (Seed), Pande, Shaqfeh (Co-coordinator); 7 graduate students, 1 postdoc.  

Affiliates:
Bao, Swope.

Thrusts

Thrust 1 - Single Polymer and Interfacial Dynamics

Thurst 2 - Interparticle Dynamics and Dynamic Cooperativity

Thrust 3 - Nanostructured Soft Materials and Nanoscale Analysis

Thrust 1 - Single Polymer and Interfacial Dynamics (top)

A major focus in this thrust is the study and control of single particle and single molecule dynamics for applications such as polymer processing and advanced sensors.  Additionally, we study the small forces and the microstructural responses that arise from flowing interfaces and that are linked to nonlinear interfacial rheology.

Figure 1. Relaxation of length in a 35c* solution of DNA demonstrating the two separated relaxation times of 2 and 105 seconds

Entangled DNA polymer dynamics in shear and extension.  Progress in a number of major areas has been made in the past year. Shaqfeh and Chu have developed the first visualization of entangled polymer (DNA) dynamics in shear flow, which may permit studies of collective polymer effects rather than isolated molecular dynamics. To accomplish this, homogeneous solutions of lambda-phage DNA were created at concentrations of 10c* through 35c*, where c* is the overlap concentration. Homogeneity was verified by confocal microscopy of solutions where a small percentage of DNA was dyed in solution. Rheological measurements demonstrated that the solutions at the highest concentrations (23 and 35c*) clearly exhibited the signature of entangled polymer solutions. The relaxation of DNA from extended configurations demonstrated two separated time scales (Figure 1) with the fast scale being nearly 20 times the Rouse time, and the long time scale being very near the theoretically predicted reptation or disengagement time. Extensional flow experiments of the same solutions demonstrate rapid chain recoil upon cessation of flow, followed by slow reptative motion over the same time scale. This work has now been extended by Dambal and Shaqfeh to include full dynamic simulation within networks of entangled slip-link chain models.

Boger fluids of high molecular weight polystyrene.  Using emulsion and microemulsion procedures, Hawker, Shaqfeh and Chu have prepared functionalized linear polymers by the copolymerization of styrene and N-hydroxysuccinimidyl acrylate. Molecular weights of up to 10,000,000 have been obtained with the active ester monomer being incorporated at up to 5 mole percent.  The active ester functional groups are still reactive and have been coupled with bodipy dye. Shaqfeh has visualized single molecule fluorescent polystyrene in tricresylphosphate and diffusion measurements have verified the theoretical hydrodynamic radius of the molecule. “Boger” fluids of the high molecular weight polystyrene in oligomeric PMMA have now been created as have cross-flow devices capable of developing a planar extensional flow and stretching these molecules.

DNA dynamics in microfluidic mixed flows.  In a Seed project, Muller is developing both cross-slot and microfluidic four-roll mill devices to accomplish flow-enhanced, rapid single molecule DNA hybridization.  These microfluidic devices will allow studies of the effects of stretching on DNA hybridization kinetics. These studies will permit rapid attachment of sequence-specific markers to create a sequence-specific sensor, provide benchmark data for extending simulation tools to include reaction kinetics, and make available an apparatus for bringing together vesicles or other supramolecular assemblies in a controlled way.   The two  microdevices, a cross-slot and a “microfluidic four-roll mill”, create stagnation point flows that allow us to trap single DNA molecules; through manipulation of the flow rates and geometry, the extension of the DNA chain and the flow type (rotation, shear, extension) can be controlled. 

Figure 2 is a schematic of both microdevices. At present the microfluidic four-roll-mill device can create mixed flows where polymer dynamics is qualitatively  different than in the “standard” extensional or shear flows. One ramification of this is polymer chain length fluctuations that
are flow-type dependent as predicted in dynamic simulations by Shaqfeh, again using fast numerical algorithms with full intramolecular hydrodynamic interactions.

Finite element simulations have been used to design the microfluidic four-roll mill that will allow flows from purely rotational to purely extensional.  This has been achieved in a high-aspect-ratio design (Figure 3), and prototypes of limited aspect ratio have been fabricated in PDMS and tested using flow visualization.

Prototypes of the cross-slot device, based on the design of Schroeder, Chu, and Shaqfeh have been fabricated using wax and PDMS soft lithography.  A flow focuser has been added to allow probes to be introduced along a stagnation streamline, and steady, reproducible flow focusing has been demonstrated.  These devices have been used to trap and stretch DNA in purely extensional flow.  In parallel, we have designed probes that allow simultaneous visualization of the DNA backbone and the probe oligonucleotide. By binding biotinylated CdSe quantum dots to streptavidin-coated polystyrene beads at varying dot to bead ratios, we can tune the brightness of the marker, which is then functionalized with the oligonucleotide probe sequence.  In bulk solution, low yield binding is observed when we hybridize these probes (an oligo complementary to a single-stranded end of ds l-DNA) to YOYO-1 labeled l-DNA (Figure 4).   The marker surface chemistry is now being optimized  for protein nucleic acid probes that can bind along internal sequences of the chain.

Stretching tethered DNA as a scaffold for organic wires.  Shaqfeh and Bao have examined a stretching flow from a surface stagnation point (Figure 5) as a means of stretching DNA between two electrodes. The surface extensional flow creates interesting dynamic properties, as predicted by numerical simulation including conformational hysteresis, which is important to understanding the engineering of the stretching flow for this application. This is the first time that conformational hysteresis has been predicted in a nonlinear flow, and in this instance the interface creates the nonlinearity that drives the phenomenon. Numerical simulations are made possible because of a new algorithm for summing long range hydrodynamic intramolecular interactions (HI) including wall HI. Large scale simulation has proven invaluable as a probe for these hysteretic or “glassy” dynamics, and these findings have significant ramifications for microfluidics of complex fluids.

Simulation of dynamics of macromolecule folding. Pande and Swope have simulated a variety of synthetic and biological systems of increasing complexity.  Understanding the folding mechanism of synthetic systems such as poly(phenylacetylene) –(PPA) may yield insight into the design of new  molecules with improved self-assembly properties and opens the door to synthetic folding analogous to biopolymers such as proteins, DNA, or RNA.  In the last two years, they have found that PPA dynamics become remarkably complex as the chain length increases.  Indeed, while their original work on PPA 12-mers found fairly simple dynamics, PPA 20-mers show a very rich, multi-state dynamical picture, with numerous traps and intermediates.  
 
Markovian state models for long-timescale simulation. One of the greatest challenges facing molecular simulation is the great divide between the time scale accessible by simulation (typically nanoseconds) and that required to match experiment (typically milliseconds to seconds). Thus in order to tackle these new goals, Pande andSwope have developed a method for long timescale kinetics.  This method is especially powerful when combined with the unique capabilities of Folding@Home, Pande’s distributed supercluster of over 160,000 CPUs.    The method Pande is developing and validating will be able to use the output of hundreds or thousands of short simulations to deduce long-time-scale behavior.  This is done by the construction of a Markov chain model for the dynamical motion.  Long-time behavior is then extracted from an eigen-analysis of the resulting Markov transition matrix.

Figure 6.  Markovian State Model methodology: trajectories are clustered and combined to create a highly connected graph of states and rates between them.

It is hoped that such models can describe not only the kinetics, but also the mechanisms and pathways (state-to-state evolution) of the folding process.  In order for this method to work, however, it is imperative that one be able to deduce an appropriate state space, with respect to which a Markov model will be appropriate.  If an inappropriate state space is used, a Markov model will not be able to describe the observed behavior – i.e., the system will appear to have history-dependent transition probabilities.  Pande has had considerable success applying this idea to a simple model problem, the alanine dipeptide (N-acetyl-Alanine-N’-methylamide).

Thrust 2 - Interparticle Dynamics and Dynamic Cooperativity (top)

We are developing and systematically studying synthetic material recognition systems based upon the principles of cooperative molecular recognition, for example as employed by living cells.  This work includes vesicles as well as particles.

 Tethered vesicles and their arrays.  One of the core goals of IRG-2 is to create systems based on supported lipid bilayers that can be used to study integral membrane proteins and multivalent interactions among membrane-associated components.  Boxer has exploited patterning methods developed earlier during CPIMA II and encoding methods based on DNA recognition to spatially array vesicles as illustrated in Figure 7.  Individual vesicles can be readily visualized by fluorescence microscopy.  Their motion is being studied as a function of surface pressure, vesicle size and charge at the air-water interface in collaboration with Fuller.  These experiments are being carried out at the surfaces of water droplets with controllable surface area and the use of micron-scale bubbles provided by Longo is planned in the near future.  Frank is following a complementary approach to tethered vesicles in which the vesicles are linked to a supported bilayer through a biotin-streptavidin-biotin coupling. The focus of the work is on the kinetics of the supramolecular assembly formation using the quartz crystal  microbalance with dissipation.  Of particular interest are shape changes in the tethered vesicles as a function of surface tethering density.

Toward the use of tethered vesicles in sensor arrays, Boxer has reported on the dynamic response of individual tethered vesicles to an electric field applied parallel to a negatively charged bilayer surface.  Vesicles respond to the field by moving in the direction of electro-osmotic flow. The electrophoretic force on negatively charged tethered vesicles opposes the electro-osmotic force.  By increasing the amount of negative charge on the tethered vesicle, drift in the direction of electro-osmotic flow can be slowed; at high negative charge on the tethered vesicle, motion can be forced in the direction of electrophoresis.  The charge gradient at the surface creates a gradient of electro-osmotic flow, and vesicles carrying similar amounts of negative charge can be focused to a region perpendicular to the applied field where electrophoresis is balanced by electro-osmosis, away from a corral boundary.  With Boxer’s tethered vesicle system, vesicle-vesicle interactions will be tracked as they are mediated by charge, complementary olignonucleotides or proteins.  It is expected that colloidal phase behavior similar to that observed by Groves for larger scale silica beads will be encountered and, thus, analysis and instrumentation developed by Groves will be utilized in this study.  This will be described shortly.

Simulation of vesicle fusion. The Pande lab has recently developed a novel technology for simulating large membrane systems and applied it to membrane fusion.  As shown in Figure 8, using a chemical crosslinker to restrain two fusogenic vesicles, he is able to observe fusion and computationally predict intermediates and kinetics in a quantitative and robust manner. Large-scale distributed computing (via the Folding@ Home distributed computing project) is used to build a Markovian Model of fusion kinetics, by sampling a series of states and the transition probabilities between these states with coarse grained Molecular Dynamics simulation. This method builds upon the work of the Pande and Swope labs by applying methods originally designed for long timescale polymer dynamics (especially the folding of proteins and non-biological foldamers) to lipid membrane dynamics.  Extensions of this model include states corresponding to the encounter complex of a pair of vesicles to quantitatively predict the pair collision and interaction dynamics seen in the Boxer lab experiments.

Figure 8: Crosslinker-induced fusion of lipid vesicles. Using worldwide distributed computing and new Markovian State Model analysis, we were able to predict the fusion of 15nm vesicles on the sub-millisecond timescale. 

In a related project, Frank has demonstrated a novel method wherein an amphipathic α-helical fusion peptide can be used to destabilize intact vesicles, transforming them into planar bilayer structure on various substrates, such as gold and titanium oxide.  Both the frequency and dissipation values from QCM-D indicate complete synthetic biomembranes in the configuration of two-dimensional complex fluids. Whereas previous researchers have been limited in their selection of surface materials for lipid biomembranes, the superior properties of gold and titanium-oxide substrates can now be utilized in various applications, such as biosensor and lab-on-a-chip devices.

Colloid-based platform for studying interfacial interactions.  Groves has recently developed several colloid-based bio-analytical assays using micron-sized silica particles coated with model membranes. 

These particles settle gravitationally into a two-dimensional dispersion where they diffuse and interact with one another eventually self-assembling into ordered or disordered structures.  The colloidal structure serves as a cooperative amplifier revealing molecular events on the membrane surface without the need for labeling or complicated experimental protocols.  An intriguing feature of this system is the appearance of an anomalous long-range attraction between like charged particles.  Groves has confirmed the existence of a long-range attraction by measuring effective particle pair interactions in low density two-dimensional colloids using a combination of inverse Monte-Carlo and Ornstein-Zernike inversion.  By changing the lipid composition, the interaction can be tuned from exclusively repulsive to having an anomalous attraction at large distance.  Although anomalous like-charge attractions have been observed previously, this is the first in which the effect has proven tunable.  The results indicate the chemistry at the surface is responsible.  Going forward, Groves will use this colloidal system and our ability to control its surface precisely to explore this anomalous attractive force. 

Interparticle interactions in two-dimensional suspension.  A project between Fuller and Shaqfeh studies two-dimensional suspensions of colloidal particles residing between oil and water. These materials can adopt a wide range of microstructures, depending on the nature of interparticle forces. Figure 10 shows three distinctly different particle arrangements, depending on the relative importance of attractive and repulsive forces. Figure 10a shows a suspension of monodisperse three micron diameter spheres.  In 10b, a bimodal mixture of one and three-micron diameter spheres is shown. The total charge on the larger spheres is larger than the charge on the smaller spheres; consequently, “holes” are created around these larger particles. Application of shear causes the holes to deform. If the larger particles are increased in size too much, however, capillary forces will draw smaller particles to their perimeter. The resulting aggregates produce quadrupolar, attractive forces and super-aggregates are formed. This is shown in 10c, where one-micron diameter spheres are shown mixed together with ten micron diameter spheres.  

Thrust 1 - Nanostructured Soft Materials and Nanoscale Analysis (top)

This thrust focuses on structures formed in polymeric and biomolecular systems that mainly exist on the nanoscale, whereas the previous thrusts studied phenomena that could often be observed on the micron-scale.  Growing out of this work is the development of nano-scale detection of compositional and topographical variations.   
 

Molecular motions at lipid bilayer interface.  Intercellular junctions create a complex and dynamic chemical environment in which collective molecular motions can be coupled to individual molecular interactions.  Groves has developed two approaches to studying the spatial and temporal correlations of molecular motion at model membrane interfaces. 

First, using real-time fluorescence interference contrast microscopy (FLIC) as shown in Figure 11, spatial and temporal spectra were monitored simultaneously to study hydrodynamic coupling between closely apposed lipid bilayer membranes.  While the membrane fluctuation timescale is strongly damped by proximity to another surface, spatial fluctuation length scales remain unaltered.  These results provide experimental confirmation of recent theoretical models and underscore the significance of confined hydrodynamic interactions on soft surfaces.  Second, mesoscopic diffusion measurements, obtained by fluorescence correlation spectroscopy (FCS) were used to characterize the effects of protein binding to the membrane surface.  Near a phase transition, long range ordering of lipids not directly involved in the protein interaction was found to increase upon protein binding.  These observations highlight the extremely responsive nature of lipid membranes near phase transitions, and illustrate the ease with which proteins may modulate such structures.

Lipid raft. It is widely believed that sphingolipid and cholesterol-enriched domains on cell membranes, sometimes called lipid rafts, are important in many biological functions.  However, how available forces and dynamics control size, structure, lifetime, mechanical properties, and mobility of these rafts is not well understood.  Through careful control of nucleation and growth conditions, as well as membrane content (e.g. addition of cholesterol and changes in lipid tail structure) Longo and Groves are studying these inter-relationships.   In particular, they have shown that supported lipid bilayers can be studied using AFM for characterization of nanometer-scale domains that are not visible using typical methods such as fluorescence microscopy as shown in Figure 12.  Groves and Longo are beginning a collaboration to interrogate the mobility of domain components contained in supported lipid bilayers utilizing FCS.

Lipid bilayer analysis with secondary ion mass spectrometry.  In order to fully implement supported bilayers in device formats, it is highly desirable to have the ability to track membrane compo- nents and bound proteins without the use of bulky fluorescent probes.  To this end, Boxer has successfully utilized high lateral resolution (50 nm) secondary ion mass spectrometry (NanoSIMS) to image isotopically labeled components in bilayers that have been freeze dried.  Longo has developed a phase-separated supported lipid bilayer composition that Boxer has used in the past year as a test system for composition analysis on the 50 nm length scale as shown in Figure 13.