Predicting response and toxicity to immune checkpoint inhibitors in lung cancer using antibodies to frameshift neoantigens.

PURPOSE: To evaluate a new class of blood-based biomarkers, anti-frameshift peptide antibodies, for predicting both tumor responses and adverse immune events to immune checkpoint inhibitor (ICI) therapies in advanced lung cancer patients. EXPERIMENTAL DESIGN: Serum samples were obtained from 74 lung cancer patients prior to palliative PD-(L)1 therapies with subsequently recorded tumor responses and immune adverse events (irAEs). Pretreatment samples were assayed on microarrays of frameshift peptides (FSPs), representing ~ 375,000 variant peptides that tumor cells can be informatically predicted to produce from translated mRNA processing errors. Serum-antibodies specifically recognizing these ligands were measured. Binding activities preferentially associated with best-response and adverse-event outcomes were determined. These antibody bound FSPs were used in iterative resampling analyses to develop predictive models of tumor response and immune toxicity. RESULTS: Lung cancer serum samples were classified based on predictive models of ICI treatment outcomes. Disease progression was predicted pretreatment with ~ 98% accuracy in the full cohort of all response categories, though ~ 30% of the samples were indeterminate. This model was built with a heterogeneous sample cohort from patients that (i) would show either clear response or stable outcomes, (ii) would be administered either single or combination therapies and (iii) were diagnosed with different lung cancer subtypes. Removing the stable disease, combination therapy or SCLC groups from model building increased the proportion of samples classified while performance remained high. Informatic analyses showed that several of the FSPs in the all-response model mapped to translations of variant mRNAs from the same genes. In the predictive model for treatment toxicities, binding to irAE-associated FSPs provided 90% accuracy pretreatment, with no indeterminates. Several of the classifying FSPs displayed sequence similarity to self-proteins. CONCLUSIONS: Anti-FSP antibodies may serve as biomarkers for predicting ICI outcomes when tested against ligands corresponding to mRNA-error derived FSPs. Model performances suggest this approach might provide a single test to predict treatment response to ICI and identify patients at high risk for immunotherapy toxicities.

Comparison of personal and shared frameshift neoantigen vaccines in a mouse mammary cancer model.

BACKGROUND: It is widely hoped that personal cancer vaccines will extend the number of patients benefiting from checkpoint and other immunotherapies. However, it is clear creating such vaccines will be challenging. It requires obtaining and sequencing tumor DNA/RNA, predicting potentially immunogenic neoepitopes and manufacturing a one-use vaccine. This process takes time and considerable cost. Importantly, most mutations will not produce an immunogenic peptide and many patient's tumors do not contain enough DNA mutations to make a vaccine. We have discovered that frameshift peptides (FSP) created from errors in the production of RNA rather than from DNA mutations are potentially a rich source of neoantigens for cancer vaccines. These errors are predictable, enabling the production of a FSP microarray. Previously we found that these microarrays can identify both personal and shared neoantigens. Here, we compared the performance of personal cancer vaccines (PCVs) with that of a shared antigen vaccine, termed Frameshift Antigen Shared Therapeutic (FAST) vaccine, using the 4 T1 breast cancer model. Sera from 4 T1-tumor bearing mice were assayed on the peptide microarray containing 200 Fs neoantigens, for the PCV, the top 10 candidates were select and personal vaccines constructed and administrated to the respective mice. For the FAST, we selected the top 10 candidates with higher prevalence among all the mice challenged. Seven to 12 days challenged mice were immunized, combined or not with immune checkpoint inhibitor (ICI) (αPD-L1 and αCTLA-4). Primary and secondary tumor clearance and growth were evaluated as well as cellular and humoral immune response against the vaccine targets by IFN-γ ELISPOT and ELISA. Lastly, we analyzed the immune response of the FAST-vaccinated mice by flow cytometry in comparison to the control group. RESULTS: We found that PCVs and FAST vaccines both reduced primary tumor incidence and growth as well as lung metastases when delivered as monotherapies or in combination with ICI. Additionally, the FAST vaccine induces a robust and effective T-cell response. CONCLUSIONS: These results suggest that FSPs produced from RNA-based errors are potent neoantigens that could enable production of off-the-shelf shared antigen vaccines for solid tumors with efficacy comparable to that of PCVs.

General Assessment of Humoral Activity in Healthy Humans.

The humoral immune system is network of biological molecules designed to maintain a healthy homeostatic equilibrium. Because antibodies are an abundant and highly specific effector of immunological action, they are also an important reservoir of previous host exposures. Antibodies may play a major role in early detection of host challenge. Unfortunately, few practical methods exist for interpreting the information stored in antibody variable regions. Immunosignatures use a microarray of thousands of random sequence peptides to interrogate antibodies in a broad and unbiased fashion. The pattern of binding between antibody and peptide is reproducible. Once the system has been trained on a disease cohort, blinded samples can be reliably predicted. Although immunosignatures of both chronic and infectious disease have been extensively tested, less has been done to demonstrate how healthy immunosignatures change over time or between individuals. Here, we report the results of a study of immunosignatures of healthy persons over brief (12 h sampled once per hour), intermediate (32 days sampled once per day), and long (5 years sampled once every year) time spans. Using this information, we were also able to detect intentional and unintentional immunological perturbations in the form of a vaccine and an infection, respectively. Our findings suggest that, even with the variability inherent in healthy immunosignatures, a single person's immunosignature will remain constant over time. Over this healthy signature, vaccines and infections create subsignatures that are common across multiple people, even subsuming healthy fluctuations. These findings have implications for disease monitoring and early diagnosis.

Cross-Reactive Synbody Affinity Ligands for Capturing Diverse Noroviruses.

Noroviruses are the most common cause of acute gastroenteritis in the developed world. Noroviruses are a diverse group of nonenveloped RNA viruses that are continuously evolving. This leads to the rise of immunologically distinct strains of the same genotype on a frequent basis. This diversity presents a unique challenge for detection and tracking of new strains, with the continuous need for new norovirus affinity ligands. Our group developed a family of bivalent synbody affinity ligands using a virus-like particle (VLP) from the 2006 GII.4 Minerva strain of norovirus. We produced more than 20 synbodies with low nanomolar dissociation constants (KD < 10 nM) for GII.4 VLP. We measured binding affinity for four synbodies against VLPs from multiple GI and GII genotypes and found that the synbodies were broadly cross-reactive with affinities that ranged from 0.5 to 8 nM. We tested the ability of these synbodies to capture norovirus from dilute solutions and found that one synbody could capture GII.4 from a 200 000-fold dilution from a norovirus positive stool sample. When these synbodies were tested for the ability to capture of multiple genotypes, we found that four different genotypes were recognized. These data demonstrate that the synbody approach can generate multiple affinity ligands for future use in norovirus detection and possible therapeutic development.

Epitope identification from fixed-complexity random-sequence peptide microarrays.

Antibodies play an important role in modern science and medicine. They are essential in many biological assays and have emerged as an important class of therapeutics. Unfortunately, current methods for mapping antibody epitopes require costly synthesis or enrichment steps, and no low-cost universal platform exists. In order to address this, we tested a random-sequence peptide microarray consisting of over 330,000 unique peptide sequences sampling 83% of all possible tetramers and 27% of pentamers. It is a single, unbiased platform that can be used in many different types of tests, it does not rely on informatic selection of peptides for a particular proteome, and it does not require iterative rounds of selection. In order to optimize the platform, we developed an algorithm that considers the significance of k-length peptide subsequences (k-mers) within selected peptides that come from the microarray. We tested eight monoclonal antibodies and seven infectious disease cohorts. The method correctly identified five of the eight monoclonal epitopes and identified both reported and unreported epitope candidates in the infectious disease cohorts. This algorithm could greatly enhance the utility of random-sequence peptide microarrays by enabling rapid epitope mapping and antigen identification.

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.

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.

Immunosignatures can predict vaccine efficacy.

The development of new vaccines would be greatly facilitated by having effective methods to predict vaccine performance. Such methods could also be helpful in monitoring individual vaccine responses to existing vaccines. We have developed "immunosignaturing" as a simple, comprehensive, chip-based method to display the antibody diversity in an individual on peptide arrays. Here we examined whether this technology could be used to develop correlates for predicting vaccine effectiveness. By using a mouse influenza infection, we show that the immunosignaturing of a natural infection can be used to discriminate a protective from nonprotective vaccine. Further, we demonstrate that an immunosignature can determine which mice receiving the same vaccine will survive. Finally, we show that the peptides comprising the correlate signatures of protection can be used to identify possible epitopes in the influenza virus proteome that are correlates of protection.

Conjugation Approach To Produce a Staphylococcus aureus Synbody with Activity in Serum.

Synbodies show promise as a new class of synthetic antibiotics. Here, we explore improvements in their activity and production through conjugation chemistry. Maleimide conjugation is a widely used conjugation strategy due to its high yield, selectivity, and low cost. We used this strategy to conjugate two antibacterial peptides to produce a bivalent antibacterial peptide, called a synbody that has bactericidal activity against methicillin resistant Staphylococcus aureus (MRSA). The synbody was prepared by conjugation of a partially d-amino acid substituted synthetic antibacterial peptide to a bis-maleimide scaffold. The synbody slowly degrades in serum, but also undergoes exchange reactions with other serum proteins, such as albumin. Therefore, we hydrolyzed the thiosuccinimide ring using a mild hydrolysis protocol to produce a new synbody with similar bactericidal activity. The synbody was now resistant to exchange reactions and maintained bactericidal activity in serum for 2 h. This work demonstrates that low-cost maleimide coupling can be used to produce antibacterial peptide conjugates with activity in serum.

The immunosignature of canine lymphoma: characterization and diagnostic application.

BACKGROUND: Cancer diagnosis in both dogs and humans is complicated by the lack of a non-invasive diagnostic test. To meet this clinical need, we apply the recently developed immunosignature assay to spontaneous canine lymphoma as clinical proof-of-concept. Here we evaluate the immunosignature as a diagnostic for spontaneous canine lymphoma at both at initial diagnosis and evaluating the disease free interval following treatment. METHODS: Sera from dogs with confirmed lymphoma (B cell n = 38, T cell n = 11) and clinically normal dogs (n = 39) were analyzed. Serum antibody responses were characterized by analyzing the binding pattern, or immunosignature, of serum antibodies on a non-natural sequence peptide microarray. Peptides were selected and tested for the ability to distinguish healthy dogs from those with lymphoma and to distinguish lymphoma subtypes based on immunophenotype. The immunosignature of dogs with lymphoma were evaluated for individual signatures. Changes in the immunosignatures were evaluated following treatment and eventual relapse. RESULTS: Despite being a clonal disease, both an individual immunosignature and a generalized lymphoma immunosignature were observed in each dog. The general lymphoma immunosignature identified in the initial set of dogs (n = 32) was able to predict disease status in an independent set of dogs (n = 42, 97% accuracy). A separate immunosignature was able to distinguish the lymphoma based on immunophenotype (n = 25, 88% accuracy). The individual immunosignature was capable of confirming remission three months following diagnosis. Immunosignature at diagnosis was able to predict which dogs with B cell lymphoma would relapse in less than 120 days (n = 33, 97% accuracy). CONCLUSION: We conclude that the immunosignature can serve as a multilevel diagnostic for canine, and potentially human, lymphoma.

Physical characterization of the "immunosignaturing effect".

Identifying new, effective biomarkers for diseases is proving to be a challenging problem. We have proposed that antibodies may offer a solution to this problem. The physical features and abundance of antibodies make them ideal biomarkers. Additionally, antibodies are often elicited early in the ontogeny of different chronic and infectious diseases. We previously reported that antibodies from patients with infectious disease and separately those with Alzheimer's disease display a characteristic and reproducible "immunosignature" on a microarray of 10,000 random sequence peptides. Here we investigate the physical and chemical parameters underlying how immunosignaturing works. We first show that a variety of monoclonal and polyclonal antibodies raised against different classes of antigens produce distinct profiles on this microarray and the relative affinities are determined. A proposal for how antibodies bind the random sequences is tested. Sera from vaccinated mice and people suffering from a fugal infection are individually assayed to determine the complexity of signals that can be distinguished. Based on these results, we propose that this simple, general and inexpensive system could be optimized to generate a new class of antibody biomarkers for a wide variety of diseases.

Design of a randomized, placebo-controlled study evaluating efficacy and safety of a cancer preventative vaccine in dogs.

Preventative anti-cancer vaccination strategies have long been hampered by the challenge of targeting the diverse array of potential tumor antigens, with successes to date limited to cancers with viral etiologies. Identification and vaccination against frameshift neoantigens conserved across multiple species and tumor histologies is a potential cancer preventative strategy currently being investigated. Companion dogs spontaneously develop cancers at a similar incidence to those in people and are a complementary comparative patient population for the development of novel anti-cancer therapeutics. In addition to an intact immune system with tumors that arise in an autochthonous tumor microenvironment, dogs also have a shorter lifespan and temporally compressed tumor natural history as compared to humans, which allows for more rapid evaluation of safety, immunogenicity, and efficacy of cancer vaccination strategies. Here we describe the study protocol for the Vaccination Against Canine Cancer Study (VACCS), the largest interventional cancer clinical trial conducted in companion dogs to date. In addition to safety and immunogenicity, the primary endpoint of VACCS is the cumulative incidence (CI) of dogs developing malignant neoplasia of any type at the end of the study period. Secondary endpoints include changes in incidence of specific tumor types, survival times following neoplasia diagnosis, and all-cause mortality.

Exploring antibody recognition of sequence space through random-sequence peptide microarrays.

A universal platform for efficiently mapping antibody epitopes would be of great use for many applications, ranging from antibody therapeutic development to vaccine design. Here we tested the feasibility of using a random peptide microarray to map antibody epitopes. Although peptide microarrays are physically constrained to ∼10(4) peptides per array, compared with 10(8) permitted in library panning approaches such as phage display, they enable a much more high though put and direct measure of binding. Long (20 mer) random sequence peptides were chosen for this study to look at an unbiased sampling of sequence space. This sampling of sequence space is sparse, as an exact epitope sequence is unlikely to appear. Commercial monoclonal antibodies with known linear epitopes or polyclonal antibodies raised against engineered 20-mer peptides were used to evaluate this array as an epitope mapping platform. Remarkably, peptides with the most sequence similarity to known epitopes were only slightly more likely to be recognized by the antibody than other random peptides. We explored the ability of two methods singly and in combination to predict the actual epitope from the random sequence peptides bound. Though the epitopes were not directly evident, subtle motifs were found among the top binding peptides for each antibody. These motifs did have some predictive ability in searching for the known epitopes among a set of decoy sequences. The second approach using a windowing alignment strategy, was able to score known epitopes of monoclonal antibodies well within the test dataset, but did not perform as well on polyclonals. Random peptide microarrays of even limited diversity may serve as a useful tool to prioritize candidates for epitope mapping or antigen identification.

Comparative study of classification algorithms for immunosignaturing data.

BACKGROUND: High-throughput technologies such as DNA, RNA, protein, antibody and peptide microarrays are often used to examine differences across drug treatments, diseases, transgenic animals, and others. Typically one trains a classification system by gathering large amounts of probe-level data, selecting informative features, and classifies test samples using a small number of features. As new microarrays are invented, classification systems that worked well for other array types may not be ideal. Expression microarrays, arguably one of the most prevalent array types, have been used for years to help develop classification algorithms. Many biological assumptions are built into classifiers that were designed for these types of data. One of the more problematic is the assumption of independence, both at the probe level and again at the biological level. Probes for RNA transcripts are designed to bind single transcripts. At the biological level, many genes have dependencies across transcriptional pathways where co-regulation of transcriptional units may make many genes appear as being completely dependent. Thus, algorithms that perform well for gene expression data may not be suitable when other technologies with different binding characteristics exist. The immunosignaturing microarray is based on complex mixtures of antibodies binding to arrays of random sequence peptides. It relies on many-to-many binding of antibodies to the random sequence peptides. Each peptide can bind multiple antibodies and each antibody can bind multiple peptides. This technology has been shown to be highly reproducible and appears promising for diagnosing a variety of disease states. However, it is not clear what is the optimal classification algorithm for analyzing this new type of data. RESULTS: We characterized several classification algorithms to analyze immunosignaturing data. We selected several datasets that range from easy to difficult to classify, from simple monoclonal binding to complex binding patterns in asthma patients. We then classified the biological samples using 17 different classification algorithms. Using a wide variety of assessment criteria, we found 'Naïve Bayes' far more useful than other widely used methods due to its simplicity, robustness, speed and accuracy. CONCLUSIONS: 'Naïve Bayes' algorithm appears to accommodate the complex patterns hidden within multilayered immunosignaturing microarray data due to its fundamental mathematical properties.

GuiTope: an application for mapping random-sequence peptides to protein sequences.

BACKGROUND: Random-sequence peptide libraries are a commonly used tool to identify novel ligands for binding antibodies, other proteins, and small molecules. It is often of interest to compare the selected peptide sequences to the natural protein binding partners to infer the exact binding site or the importance of particular residues. The ability to search a set of sequences for similarity to a set of peptides may sometimes enable the prediction of an antibody epitope or a novel binding partner. We have developed a software application designed specifically for this task. RESULTS: GuiTope provides a graphical user interface for aligning peptide sequences to protein sequences. All alignment parameters are accessible to the user including the ability to specify the amino acid frequency in the peptide library; these frequencies often differ significantly from those assumed by popular alignment programs. It also includes a novel feature to align di-peptide inversions, which we have found improves the accuracy of antibody epitope prediction from peptide microarray data and shows utility in analyzing phage display datasets. Finally, GuiTope can randomly select peptides from a given library to estimate a null distribution of scores and calculate statistical significance. CONCLUSIONS: GuiTope provides a convenient method for comparing selected peptide sequences to protein sequences, including flexible alignment parameters, novel alignment features, ability to search a database, and statistical significance of results. The software is available as an executable (for PC) at http://www.immunosignature.com/software and ongoing updates and source code will be available at sourceforge.net.

Application of immunosignatures to the assessment of Alzheimer's disease.

OBJECTIVE: Accurate assessment of Alzheimer's disease (AD), both presymptomatically and at different disease stages, will become increasingly important with the expanding elderly population. There are a number of indications that the immune system is engaged in AD. Here we explore the ability of an antibody-profiling technology to characterize AD and screen for peptides that may be used for a simple diagnostic test. METHODS: We developed an array-based system to profile the antibody repertoire of transgenic mice with cerebral amyloidosis (TG) and elderly individuals with or without AD. The array consists of 10,000 random sequence peptides (20-mers) capable of detecting antibody binding patterns, allowing the identification of peptides that mimic epitopes targeted by a donor's serum. RESULTS: TG mice exhibited a distinct immunoprofile compared to nontransgenic littermates. Further, we show that dementia patients, including autopsy-confirmed AD subjects, have distinguishable profiles compared to age-matched nondemented people. Using antibodies to different forms of Aß peptide and blocking protocols, we demonstrate that most of this signature is not due to the subject's antibodies raised against Aß. INTERPRETATION: We propose that "immunosignaturing" technology may have potential as a diagnostic tool in AD.

Autoreactive antibodies raised by self derived de novo peptides can identify unrelated antigens on protein microarrays. Are autoantibodies really autoantibodies?

The development of arrays of human proteins has been a huge boon to the search for autoantibody diagnostics. Typically, slides with thousands of recombinant human proteins arrayed in an addressable fashion are incubated with sera from diseased or normal people. If an antibody binds a protein more in the diseased than in the normal cohort it is considered an autoantibody response. It is usually presumed that the autoantibody was elicited by the protein bound on the array. However, our studies using human protein and random peptide arrays indicate that antibody specificity may not be as high as commonly thought. Therefore we have tested the assumption of the source of autoantibodies. One test was to generate antibodies to two totally random peptides and bind these antibodies to a human protein array. One of the antibodies generated bound two human proteins. A second test was to generate an antibody to a frameshift peptide occurring in cancers. This antibody also bound several proteins on the array. We conclude that one should be cautious about assuming a particular autoantibody target on an array which elicited the original immune response.

Proteolytic turnover of the Gal4 transcription factor is not required for function in vivo.

Transactivator-promoter complexes are essential intermediates in the activation of eukaryotic gene expression. Recent studies of these complexes have shown that some are quite dynamic in living cells owing to rapid and reversible disruption of activator-promoter complexes by molecular chaperones, or a slower, ubiquitin-proteasome-pathway-mediated turnover of DNA-bound activator. These mechanisms may act to ensure continued responsiveness of activators to signalling cascades by limiting the lifetime of the active protein-DNA complex. Furthermore, the potency of some activators is compromised by proteasome inhibition, leading to the suggestion that periodic clearance of activators from a promoter is essential for high-level expression. Here we describe a variant of the chromatin immunoprecipitation assay that has allowed direct observation of the kinetic stability of native Gal4-promoter complexes in yeast. Under non-inducing conditions, the complex is dynamic, but on induction the Gal4-promoter complexes 'lock in' and exhibit long half-lives. Inhibition of proteasome-mediated proteolysis had little or no effect on Gal4-mediated gene expression. These studies, combined with earlier data, show that the lifetimes of different transactivator-promoter complexes in vivo can vary widely and that proteasome-mediated turnover is not a general requirement for transactivator function.

Engineering a synthetic ligand for tumor necrosis factor-alpha.

One approach to prepare protein binding ligands is to join two low-affinity ligands that bind different sites on the target protein to create a high-affinity bivalent ligand. This typically requires some knowledge of the ligand binding site and requires exquisite orientation of the ligands in order to achieve maximum binding affinity. Here, we explored the limit of affinity improvement possible with no a priori knowledge of peptide binding site and with minimal effort spent in linking the lead peptides. We compared the affinity enhancement from linking two peptides with low affinity for tumor necrosis factor-α (TNFA) to the affinity enhancement from linking affinity improved versions of these peptides using several different scaffolds. We found that we achieved the highest affinity gain not by the precise positioning of the peptides, but rather by using affinity improved versions of the lead peptides to produce synbodies with apparent K(D)'s of 9 to 48 nM. Kinetic analysis showed that the binding kinetics of the synbody are strongly influenced by the kinetics of the starting peptide. This suggests that careful selection of peptides based on their kinetic profile prior to linking will influence the kinetics of the final binding agent.