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.

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.

Production of high-complexity frameshift neoantigen peptide microarrays.

Parallel measurement of large numbers of antigen-antibody interactions are increasingly enabled by peptide microarray technologies. Our group has developed an in situ synthesized peptide microarray of >400 000 frameshift neoantigens using mask-based photolithographic peptide synthesis, to profile patient specific neoantigen reactive antibodies in a single assay. The system produces 208 replicate mircoarrays per wafer and is capable of producing multiple wafers per synthetic lot to routinely synthesize over 300 million peptides simultaneously. In this report, we demonstrate the feasibility of the system for detecting peripheral-blood antibody binding to frameshift neoantigens across multiple synthetic lots.

Peptide Sequencing Directly on Solid Surfaces Using MALDI Mass Spectrometry.

There are an increasing variety of applications in which peptides are both synthesized and used attached to solid surfaces. This has created a need for high throughput sequence analysis directly on surfaces. However, common sequencing approaches that can be adapted to surface bound peptides lack the throughput often needed in library-based applications. Here we describe a simple approach for sequence analysis directly on solid surfaces that is both high speed and high throughput, utilizing equipment available in most protein analysis facilities. In this approach, surface bound peptides, selectively labeled at their N-termini with a positive charge-bearing group, are subjected to controlled degradation in ammonia gas, resulting in a set of fragments differing by a single amino acid that remain spatially confined on the surface they were bound to. These fragments can then be analyzed by MALDI mass spectrometry, and the peptide sequences read directly from the resulting spectra.

A Simple Platform for the Rapid Development of Antimicrobials.

Recent infectious outbreaks highlight the need for platform technologies that can be quickly deployed to develop therapeutics needed to contain the outbreak. We present a simple concept for rapid development of new antimicrobials. The goal was to produce in as little as one week thousands of doses of an intervention for a new pathogen. We tested the feasibility of a system based on antimicrobial synbodies. The system involves creating an array of 100 peptides that have been selected for broad capability to bind and/or kill viruses and bacteria. The peptides are pre-screened for low cell toxicity prior to large scale synthesis. Any pathogen is then assayed on the chip to find peptides that bind or kill it. Peptides are combined in pairs as synbodies and further screened for activity and toxicity. The lead synbody can be quickly produced in large scale, with completion of the entire process in one week.

Whole-Virus Screening to Develop Synbodies for the Influenza Virus.

There is an ongoing need for affinity agents for emerging viruses and new strains of current human viruses. We therefore developed a robust and modular system for engineering high-affinity synbody ligands for the influenza A/Puerto Rico/8/1934 H1N1 virus as a model system. Whole-virus screening against a peptide microarray was used to identify binding peptides. Candidate peptides were linked to bis-maleimide peptide scaffolds to produce a library of candidate influenza-binding synbodies. From this library, a candidate synbody, ASU1060, was selected and affinity-improved via positional substitution using d-amino acids to produce a new synbody, ASU1061, that bound H1N1 in an ELISA assay with a KD of <1 nM, comparable to that of a monoclonal antibody for neuraminidase (NA). We prepared a modified version of ASU1061 that contained an additional C-terminal peptide to simulate conjugation of the synbody to a carrier protein, called ASU1063, and found that H1N1 binding was unchanged. Subsequent work identified the synbody target as nucleoprotein (NP), a highly conserved protein in influenza, with a KD of <1 nM for ASU1063. This suggests that virus-binding synbodies can be conjugated to carrier proteins or other moieties that could improve the therapeutic profile of the resulting synbody. This method is a rapid process that offers a means of developing new affinity ligands to influenza and other viruses.

Scalable high-density peptide arrays for comprehensive health monitoring.

There is an increasing awareness that health care must move from post-symptomatic treatment to presymptomatic intervention. An ideal system would allow regular inexpensive monitoring of health status using circulating antibodies to report on health fluctuations. Recently, we demonstrated that peptide microarrays can do this through antibody signatures (immunosignatures). Unfortunately, printed microarrays are not scalable. Here we demonstrate a platform based on fabricating microarrays (~10 M peptides per slide, 330,000 peptides per assay) on silicon wafers using equipment common to semiconductor manufacturing. The potential of these microarrays for comprehensive health monitoring is verified through the simultaneous detection and classification of six different infectious diseases and six different cancers. Besides diagnostics, these high-density peptide chips have numerous other applications both in health care and elsewhere.

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.