Modeling the sequence dependence of differential antibody binding in the immune response to infectious disease.

Past studies have shown that incubation of human serum samples on high density peptide arrays followed by measurement of total antibody bound to each peptide sequence allows detection and discrimination of humoral immune responses to a variety of infectious diseases. This is true even though these arrays consist of peptides with near-random amino acid sequences that were not designed to mimic biological antigens. This "immunosignature" approach, is based on a statistical evaluation of the binding pattern for each sample but it ignores the information contained in the amino acid sequences that the antibodies are binding to. Here, similar array-based antibody profiles are instead used to train a neural network to model the sequence dependence of molecular recognition involved in the immune response of each sample. The binding profiles used resulted from incubating serum from 5 infectious disease cohorts (Hepatitis B and C, Dengue Fever, West Nile Virus and Chagas disease) and an uninfected cohort with 122,926 peptide sequences on an array. These sequences were selected quasi-randomly to represent an even but sparse sample of the entire possible combinatorial sequence space (~1012). This very sparse sampling of combinatorial sequence space was sufficient to capture a statistically accurate representation of the humoral immune response across the entire space. Processing array data using the neural network not only captures the disease-specific sequence-binding information but aggregates binding information with respect to sequence, removing sequence-independent noise and improving the accuracy of array-based classification of disease compared with the raw binding data. Because the neural network model is trained on all samples simultaneously, a highly condensed representation of the differential information between samples resides in the output layer of the model, and the column vectors from this layer can be used to represent each sample for classification or unsupervised clustering applications.

Two-Dimensional Excitonic Networks Directed by DNA Templates as an Efficient Model Light-Harvesting and Energy Transfer System.

Photosynthetic organisms organize discrete light-harvesting complexes into large-scale networks to facilitate efficient light collection and utilization. Inspired by nature, herein, synthetic DNA templates were used to direct the formation of dye aggregates with a cyanine dye, K21, into discrete branched photonic complexes, and two-dimensional (2D) excitonic networks. The DNA templates ranged from four-arm DNA tiles, ≈10 nm in each arm, to 2D wireframe DNA origami nanostructures with different geometries and varying dimensions up to 100×100 nm. These DNA-templated dye aggregates presented strongly coupled spectral features and delocalized exciton characteristics, enabling efficient photon collection and energy transfer. Compared to the discrete branched photonic systems templated on individual DNA tiles, the interconnected excitonic networks showed approximately a 2-fold increase in energy transfer efficiency. This bottom-up assembly strategy paves the way to create 2D excitonic systems with complex geometries and engineered energy pathways.

Efficient Long-Range, Directional Energy Transfer through DNA-Templated Dye Aggregates.

The benzothiazole cyanine dye K21 forms dye aggregates on double-stranded DNA (dsDNA) templates. These aggregates exhibit a red-shifted absorption band, enhanced fluorescence emission, and an increased fluorescence lifetime, all indicating strong excitonic coupling among the dye molecules. K21 aggregate formation on dsDNA is only weakly sequence dependent, providing a flexible approach that is adaptable to many different DNA nanostructures. Donor (D)-bridge (B)-acceptor (A) complexes consisting of Alexa Fluor 350 as the donor, a 30 bp (9.7 nm) DNA templated K21 aggregate as the bridge, and Alexa Fluor 555 as the acceptor show an overall donor to acceptor energy transfer efficiency of ∼60%, with the loss of excitation energy being almost exclusively at the donor-bridge junction (63%). There was almost no excitation energy loss due to transfer through the aggregate bridge, and the transfer efficiency from the aggregate to the acceptor was about 96%. By comparing the energy transfer in templated aggregates at several lengths up to 32 nm, the loss of energy per nanometer through the K21 aggregate bridge was determined to be <1%, suggesting that it should be possible to construct structures that use much longer energy transfer "wires" for light-harvesting applications in photonic systems.

Programmed coherent coupling in a synthetic DNA-based excitonic circuit.

Natural light-harvesting systems spatially organize densely packed chromophore aggregates using rigid protein scaffolds to achieve highly efficient, directed energy transfer. Here, we report a synthetic strategy using rigid DNA scaffolds to similarly program the spatial organization of densely packed, discrete clusters of cyanine dye aggregates with tunable absorption spectra and strongly coupled exciton dynamics present in natural light-harvesting systems. We first characterize the range of dye-aggregate sizes that can be templated spatially by A-tracts of B-form DNA while retaining coherent energy transfer. We then use structure-based modelling and quantum dynamics to guide the rational design of higher-order synthetic circuits consisting of multiple discrete dye aggregates within a DX-tile. These programmed circuits exhibit excitonic transport properties with prominent circular dichroism, superradiance, and fast delocalized exciton transfer, consistent with our quantum dynamics predictions. This bottom-up strategy offers a versatile approach to the rational design of strongly coupled excitonic circuits using spatially organized dye aggregates for use in coherent nanoscale energy transport, artificial light-harvesting, and nanophotonics.

Orange Carotenoid Protein as a Control Element in an Antenna System Based on a DNA Nanostructure.

Taking inspiration from photosynthetic mechanisms in natural systems, we introduced a light-sensitive photo protective quenching element to an artificial light-harvesting antenna model to control the flow of energy as a function of light intensity excitation. The orange carotenoid protein (OCP) is a nonphotochemical quencher in cyanobacteria: under high-light conditions, the protein undergoes a spectral shift, and by binding to the phycobilisome, it absorbs excess light and dissipates it as heat. By the use of DNA as a scaffold, an antenna system made of organic dyes (Cy3 and Cy5) was constructed, and OCP was assembled on it as a modulated quenching element. By controlling the illumination intensity, it is possible to switch the direction of excitation energy transfer from the donor Cy3 to either of two acceptors. Under low-light conditions, energy is transferred from Cy3 to Cy5, and under intense illumination, energy is partially transferred to OCP as well. These results demonstrate the feasibility of controlling the pathway of energy transfer using light intensity in an engineered light-harvesting system.

Immunosignature system for diagnosis of cancer.

Although the search for disease biomarkers continues, the clinical return has thus far been disappointing. The complexity of the body's response to disease makes it difficult to represent this response with only a few biomarkers, particularly when many are present at low levels. An alternative to the typical reductionist biomarker paradigm is an assay we call an "immunosignature." This approach leverages the response of antibodies to disease-related changes, as well as the inherent signal amplification associated with antigen-stimulated B-cell proliferation. To perform an immunosignature assay, the antibodies in diluted blood are incubated with a microarray of thousands of random sequence peptides. The pattern of binding to these peptides is the immunosignature. Because the peptide sequences are completely random, the assay is effectively disease-agnostic, potentially providing a comprehensive diagnostic on multiple diseases simultaneously. To explore the ability of an immunosignature to detect and identify multiple diseases simultaneously, 20 samples from each of five cancer cohorts collected from multiple sites and 20 noncancer samples (120 total) were used as a training set to develop a reference immunosignature. A blinded evaluation of 120 blinded samples covering the same diseases gave 95% classification accuracy. To investigate the breadth of the approach and test sensitivity to biological diversity further, immunosignatures of >1,500 historical samples comprising 14 different diseases were examined by training with 75% of the samples and testing the remaining 25%. The average accuracy was >98%. These results demonstrate the potential power of the immunosignature approach in the accurate, simultaneous classification of disease.

Electron Transfer in Bacterial Reaction Centers with the Photoactive Bacteriopheophytin Replaced by a Bacteriochlorophyll through Coordinating Ligand Substitution.

The influence of amino acid substitutions at position M214 (M-subunit, residue 214) on the rate and pathway of electron transfer involving the bacteriopheophytin cofactor, HA, in a bacterial photosynthetic reaction center has been explored in a series of Rhodobacter sphaeroides mutants. The M214 leucine (L) residue of the wild type was replaced with histidine (H), glutamine (Q), and asparagine (N), creating the mutants M214LH, M214LQ, and M214LN, respectively. As has been reported previously for M214LH, each of these mutations resulted in a bacteriochlorophyll molecule in place of a bacteriopheophytin in the HA pocket, forming so-called ß-type mutants (in which the HA cofactor is called ßA). In addition, these mutations changed the properties of the surrounding protein environment in terms of charge distribution and the amino acid side chain volume. Electron transfer reactions from the excited primary donor P to the acceptor QA were characterized using ultrafast transient absorption spectroscopic techniques. Similar to that of the previously characterized M214LH (ß mutant), the strong energetic mixing of the P(+)BA(-) and P(+)ßA(-) states (the mixed anion is denoted I(-)) increased the rate of charge recombination between P(+) and I(-) in competition with the I(-) → QA forward reaction. This reduced the overall yield of charge separation forming the P(+)QA(-) state. While the kinetics of the primary electron transfer forming P(+)I(-) were essentially identical in all three ß mutants, the rates of the ßA(-) (I(-)) → QA electron transfer in M214LQ and M214LH were very similar but quite different from that of the M214LN mutant. The observed yield changes and the differences in kinetics are correlated more closely with the volume of the mutated amino acid than with their charge characteristics. These results are consistent with those of previous studies of a series of M214 mutants with different sizes of amino acid side chains that did not alter the HA cofactor composition [Pan, J., et al. (2013) J. Phys. Chem. B 117, 7179-7189]. Both studies indicate that protein relaxation in this region of the reaction center plays a key role in stabilizing charge-separated states involving the HA or ßA cofactor. The effect is particularly pronounced for reactions occurring on time scales of tens and hundreds of picoseconds (forward transfer to the QA and charge recombination).

A Three-Enzyme Pathway with an Optimised Geometric Arrangement to Facilitate Substrate Transfer.

Cascade reactions drive and regulate a variety of metabolic activities. Efficient coupling of substrate transport between enzymes is important for overall pathway activity and also controls the depletion of intermediate molecules that drive the reaction forward. Here, we assembled a three-enzyme pathway on a series of DNA nanoscaffolds to investigate the dependence of their activities on spatial arrangement. Unlike previous studies, the overall activity of the three-enzyme pathway relied less on inter-enzyme distance and more on the geometric patterns that arranged them within a relatively small range of 10-30 nm. Pathway intermediate detection demonstrated that the assembled enzyme systems quickly depleted the intermediate molecules through efficient reaction coupling.

Structural and kinetic properties of Rhodobacter sphaeroides photosynthetic reaction centers containing exclusively Zn-coordinated bacteriochlorophyll as bacteriochlorin cofactors.

The Zn-BChl-containing reaction center (RC) produced in a bchD (magnesium chelatase) mutant of Rhodobacter sphaeroides assembles with six Zn-bacteriochlorophylls (Zn-BChls) in place of four Mg-containing bacteriochlorophylls (BChls) and two bacteriopheophytins (BPhes). This protein presents unique opportunities for studying biological electron transfer, as Zn-containing chlorins can exist in 4-, 5-, and (theoretically) 6-coordinate states within the RC. In this paper, the electron transfer perturbations attributed exclusively to coordination state effects are separated from those attributed to the presence, absence, or type of metal in the bacteriochlorin at the HA pocket of the RC. The presence of a 4-coordinate Zn(2+) ion in the HA bacteriochlorin instead of BPhe results in a small decrease in the rates of the P*→P(+)HA(-)→P(+)QA(-) electron transfer, and the charge separation yield is not greatly perturbed; however coordination of the Zn(2+) by a fifth ligand provided by a histidine residue results in a larger rate decrease and yield loss. We also report the first crystal structure of a Zn-BChl-containing RC, confirming that the HA Zn-BChl was either 4- or 5-coordinate in the two types of Zn-BChl-containing RCs studied here. Interestingly, a large degree of disorder, in combination with a relatively weak anomalous difference electron density was found in the HB pocket. These data, in combination with spectroscopic results, indicate partial occupancy of this binding pocket. These findings provide insights into the use of BPhe as the bacteriochlorin pigment of choice at HA in both BChl- and Zn-BChl-containing RCs found in nature.

Reengineering the optical absorption cross-section of photosynthetic reaction centers.

Engineered cysteine residues near the primary electron donor (P) of the reaction center from the purple photosynthetic bacterium Rhodobacter sphaeroides were covalently conjugated to each of several dye molecules in order to explore the geometric design and spectral requirements for energy transfer between an artificial antenna system and the reaction center. An average of 2.5 fluorescent dye molecules were attached at specific locations near P. The enhanced absorbance cross-section afforded by conjugation of Alexa Fluor 660 dyes resulted in a 2.2-fold increase in the formation of reaction center charge-separated state upon intensity-limited excitation at 650 nm. The effective increase in absorbance cross-section resulting from the conjugation of two other dyes, Alexa Fluor 647 and Alexa Fluor 750, was also investigated. The key parameters that dictate the efficiency of dye-to-reaction center energy transfer and subsequent charge separation were examined using both steady-state and time-resolved fluorescence spectroscopy as well as transient absorbance spectroscopy techniques. An understanding of these parameters is an important first step toward developing more complex model light-harvesting systems integrated with reaction centers.

A DNA-directed light-harvesting/reaction center system.

A structurally and compositionally well-defined and spectrally tunable artificial light-harvesting system has been constructed in which multiple organic dyes attached to a three-arm-DNA nanostructure serve as an antenna conjugated to a photosynthetic reaction center isolated from Rhodobacter sphaeroides 2.4.1. The light energy absorbed by the dye molecules is transferred to the reaction center, where charge separation takes place. The average number of DNA three-arm junctions per reaction center was tuned from 0.75 to 2.35. This DNA-templated multichromophore system serves as a modular light-harvesting antenna that is capable of being optimized for its spectral properties, energy transfer efficiency, and photostability, allowing one to adjust both the size and spectrum of the resulting structures. This may serve as a useful test bed for developing nanostructured photonic systems.

Predicting monoclonal antibody binding sequences from a sparse sampling of all possible sequences.

Previous work has shown that binding of target proteins to a sparse, unbiased sample of all possible peptide sequences is sufficient to train a machine learning model that can then predict, with statistically high accuracy, target binding to any possible peptide sequence of similar length. Here, highly sequence-specific molecular recognition is explored by measuring binding of 8 monoclonal antibodies (mAbs) with specific linear cognate epitopes to an array containing 121,715 near-random sequences about 10 residues in length. Network models trained on resulting sequence-binding values are used to predict the binding of each mAb to its cognate sequence and to an in silico generated one million random sequences. The model always ranks the binding of the cognate sequence in the top 100 sequences, and for 6 of the 8 mAbs, the cognate sequence ranks in the top ten. Practically, this approach has potential utility in selecting highly specific mAbs for therapeutics or diagnostics. More fundamentally, this demonstrates that very sparse random sampling of a large amino acid sequence spaces is sufficient to generate comprehensive models predictive of highly specific molecular recognition.

Protein dielectric environment modulates the electron-transfer pathway in photosynthetic reaction centers.

The replacement of tyrosine by aspartic acid at position M210 in the photosynthetic reaction center of Rhodobacter sphaeroides results in the generation of a fast charge recombination pathway that is not observed in the wild-type. Apparently, the initially formed charge-separated state (cation of the special pair, P, and anion of the A-side bacteriopheophytin, H(A)) can decay rapidly via recombination through the neighboring bacteriochlorophyll (B(A)) soon after formation. The charge-separated state then relaxes over tens of picoseconds and recombination slows to the hundreds-of-picoseconds or nanosecond timescale. This dielectric relaxation results in a time-dependent blue shift of B(A)(-) absorption, which can be monitored using transient absorbance measurements. Protein dynamics also appear to modulate the electron transfer between H(A) and the next electron carrier, Q(A) (a ubiquinone). The kinetics of this reaction are complex in the mutant, requiring two kinetic terms, and the spectra associated with the two terms are distinct; a red shift of the H(A) ground-state bleaching is observed between the shorter and longer H(A)-to-Q(A) electron-transfer phases. The kinetics appears to be pH-independent, suggesting a negligible contribution of static heterogeneity originating from protonation/deprotonation in the ground state. A dynamic model based on the energy levels of the two early charge-separated states, P(+)B(A)(-) and P(+)H(A)(-), has been developed in which the energetics of these states is modulated by fast protein dielectric relaxations and this in turn alters both the kinetic complexity of the reaction and the reaction pathway.

Interenzyme substrate diffusion for an enzyme cascade organized on spatially addressable DNA nanostructures.

Spatially addressable DNA nanostructures facilitate the self-assembly of heterogeneous elements with precisely controlled patterns. Here we organized discrete glucose oxidase (GOx)/horseradish peroxidase (HRP) enzyme pairs on specific DNA origami tiles with controlled interenzyme spacing and position. The distance between enzymes was systematically varied from 10 to 65 nm, and the corresponding activities were evaluated. The study revealed two different distance-dependent kinetic processes associated with the assembled enzyme pairs. Strongly enhanced activity was observed for those assemblies in which the enzymes were closely spaced, while the activity dropped dramatically for enzymes as little as 20 nm apart. Increasing the spacing further resulted in a much weaker distance dependence. Combined with diffusion modeling, the results suggest that Brownian diffusion of intermediates in solution governed the variations in activity for more distant enzyme pairs, while dimensionally limited diffusion of intermediates across connected protein surfaces contributed to the enhancement in activity for closely spaced GOx/HRP assemblies. To further test the role of limited dimensional diffusion along protein surfaces, a noncatalytic protein bridge was inserted between GOx and HRP to connect their hydration shells. This resulted in substantially enhanced activity of the enzyme pair.

Electron transfer in the Rhodobacter sphaeroides reaction center assembled with zinc bacteriochlorophyll.

The cofactor composition and electron-transfer kinetics of the reaction center (RC) from a magnesium chelatase (bchD) mutant of Rhodobacter sphaeroides were characterized. In this RC, the special pair (P) and accessory (B) bacteriochlorophyll (BChl) -binding sites contain Zn-BChl rather than BChl a. Spectroscopic measurements reveal that Zn-BChl also occupies the H sites that are normally occupied by bacteriopheophytin in wild type, and at least 1 of these Zn-BChl molecules is involved in electron transfer in intact Zn-RCs with an efficiency of >95% of the wild-type RC. The absorption spectrum of this Zn-containing RC in the near-infrared region associated with P and B is shifted from 865 to 855 nm and from 802 to 794 nm respectively, compared with wild type. The bands of P and B in the visible region are centered at 600 nm, similar to those of wild type, whereas the H-cofactors have a band at 560 nm, which is a spectral signature of monomeric Zn-BChl in organic solvent. The Zn-BChl H-cofactor spectral differences compared with the P and B positions in the visible region are proposed to be due to a difference in the 5th ligand coordinating the Zn. We suggest that this coordination is a key feature of protein-cofactor interactions, which significantly contributes to the redox midpoint potential of H and the formation of the charge-separated state, and provides a unifying explanation for the properties of the primary acceptor in photosystems I (PS1) and II (PS2).

Exploring peptide space for enzyme modulators.

A method is presented for screening high-density arrays to discover peptides that bind and modulate enzyme activity. A polyvinyl alcohol solution was applied to array surfaces to limit the diffusion of product molecules released from enzymatic reactions, allowing the simultaneous measurement of enzyme activity and binding at each peptide spot. For proof of concept, it was possible to identify peptides that bound to horseradish peroxidase, alkaline phosphatase, and beta-galactosidase and substantially altered enzyme activity by comparing the binding level of peptide to enzyme and bound enzyme activity. This basic technique may be generally applicable to find peptides or other small molecules that modify enzyme activity.

High-throughput screening in two dimensions: binding intensity and off-rate on a peptide microarray.

We report a high-throughput two-dimensional microarray-based screen, incorporating both target binding intensity and off-rate, which can be used to analyze thousands of compounds in a single binding assay. Relative binding intensities and time-resolved dissociation are measured for labeled tumor necrosis factor alpha (TNF-alpha) bound to a peptide microarray. The time-resolved dissociation is fitted to a one-component exponential decay model, from which relative dissociation rates are determined for all peptides with binding intensities above background. We show that most peptides with the slowest off-rates on the microarray also have the slowest off-rates when measured by surface plasmon resonance (SPR).

Feature-level MALDI-MS characterization of in situ-synthesized peptide microarrays.

Characterizing the chemical composition of microarray features is a difficult yet important task in the production of in situ-synthesized microarrays. Here, we describe a method to determine the chemical composition of microarray features, directly on the feature. This method utilizes nondiffusional chemical cleavage from the surface along with techniques from MALDI-MS tissue imaging, thereby making the chemical characterization of high-density microarray features simple, accurate, and amenable to high-throughput.

Comprehensive Prediction of Molecular Recognition in a Combinatorial Chemical Space Using Machine Learning.

In combinatorial chemical approaches, optimizing the composition and arrangement of building blocks toward a particular function has been done using a number of methods, including high throughput molecular screening, molecular evolution, and computational prescreening. Here, a different approach is considered that uses sparse measurements of library molecules as the input to a machine learning algorithm which generates a comprehensive, quantitative relationship between covalent molecular structure and function that can then be used to predict the function of any molecule in the possible combinatorial space. To test the feasibility of the approach, a defined combinatorial chemical space consisting of ∼1012 possible linear combinations of 16 different amino acids was used. The binding of a very sparse, but nearly random, sampling of this amino acid sequence space to 9 different protein targets is measured and used to generate a general relationship between peptide sequence and binding for each target. Surprisingly, measuring as little as a few hundred to a few thousand of the ∼1012 possible molecules provides sufficient training to be highly predictive of the binding of the remaining molecules in the combinatorial space. Furthermore, measuring only amino acid sequences that bind weakly to a target allows the accurate prediction of which sequences will bind 10-100 times more strongly. Thus, the molecular recognition information contained in a tiny fraction of molecules in this combinatorial space is sufficient to characterize any set of molecules randomly selected from the entire space, a fact that potentially has significant implications for the design of new chemical function using combinatorial chemical libraries.

Creating protein affinity reagents by combining peptide ligands on synthetic DNA scaffolds.

A full understanding of the proteome will require ligands to all of the proteins encoded by genomes. While antibodies represent the principle affinity reagents used to bind proteins, their limitations have created a need for new ligands to large numbers of proteins. Here we propose a general concept to obtain protein affinity reagents that avoids animal immunization and iterative selection steps. Central to this process is the idea that small peptide libraries contain sequences that will bind to independent regions on a protein surface and that these ligands can be combined on synthetic scaffolds to create high affinity bivalent reagents. To demonstrate the feasibility of this approach, an array of 4000 unique 12-mer peptides was screened to identify sequences that bind to nonoverlapping sites on the yeast regulatory protein Gal80. Individual peptide ligands were screened at different distances using a novel DNA linking strategy to identify the optimal peptide pair and peptide pair separation distance required to transform two weaker ligands into a single high affinity protein capture reagent. A synthetic antibody or synbody was created with 5 nM affinity to Gal80 that functions in conventional ELISA and pull-down assays. We validated our synthetic antibody approach by creating a second synbody to human transferrin. In both cases, we observed an increase in binding affinity of approximately 1000-fold (DeltaDeltaG = approximately 4.1 kcal/mol) between the individual peptides and final bivalent synbody construct.