RNA vaccines for cancer: Principles to practice.

Vaccines are the most impactful medicines to improve health. Though potent against pathogens, vaccines for cancer remain an unfulfilled promise. However, recent advances in RNA technology coupled with scientific and clinical breakthroughs have spurred rapid discovery and potent delivery of tumor antigens at speed and scale, transforming cancer vaccines into a tantalizing prospect. Yet, despite being at a pivotal juncture, with several randomized clinical trials maturing in upcoming years, several critical questions remain: which antigens, tumors, platforms, and hosts can trigger potent immunity with clinical impact? Here, we address these questions with a principled framework of cancer vaccination from antigen detection to delivery. With this framework, we outline features of emergent RNA technology that enable rapid, robust, real-time vaccination with somatic mutation-derived neoantigens-an emerging "ideal" antigen class-and highlight latent features that have sparked the belief that RNA could realize the enduring vision for vaccines against cancer.

Identification of transcriptional programs using dense vector representations defined by mutual information with GeneVector.

Deciphering individual cell phenotypes from cell-specific transcriptional processes requires high dimensional single cell RNA sequencing. However, current dimensionality reduction methods aggregate sparse gene information across cells, without directly measuring the relationships that exist between genes. By performing dimensionality reduction with respect to gene co-expression, low-dimensional features can model these gene-specific relationships and leverage shared signal to overcome sparsity. We describe GeneVector, a scalable framework for dimensionality reduction implemented as a vector space model using mutual information between gene expression. Unlike other methods, including principal component analysis and variational autoencoders, GeneVector uses latent space arithmetic in a lower dimensional gene embedding to identify transcriptional programs and classify cell types. In this work, we show in four single cell RNA-seq datasets that GeneVector was able to capture phenotype-specific pathways, perform batch effect correction, interactively annotate cell types, and identify pathway variation with treatment over time.

Personalized RNA neoantigen vaccines stimulate T cells in pancreatic cancer.

Pancreatic ductal adenocarcinoma (PDAC) is lethal in 88% of patients1, yet harbours mutation-derived T cell neoantigens that are suitable for vaccines 2,3. Here in a phase I trial of adjuvant autogene cevumeran, an individualized neoantigen vaccine based on uridine mRNA-lipoplex nanoparticles, we synthesized mRNA neoantigen vaccines in real time from surgically resected PDAC tumours. After surgery, we sequentially administered atezolizumab (an anti-PD-L1 immunotherapy), autogene cevumeran (a maximum of 20 neoantigens per patient) and a modified version of a four-drug chemotherapy regimen (mFOLFIRINOX, comprising folinic acid, fluorouracil, irinotecan and oxaliplatin). The end points included vaccine-induced neoantigen-specific T cells by high-threshold assays, 18-month recurrence-free survival and oncologic feasibility. We treated 16 patients with atezolizumab and autogene cevumeran, then 15 patients with mFOLFIRINOX. Autogene cevumeran was administered within 3 days of benchmarked times, was tolerable and induced de novo high-magnitude neoantigen-specific T cells in 8 out of 16 patients, with half targeting more than one vaccine neoantigen. Using a new mathematical strategy to track T cell clones (CloneTrack) and functional assays, we found that vaccine-expanded T cells comprised up to 10% of all blood T cells, re-expanded with a vaccine booster and included long-lived polyfunctional neoantigen-specific effector CD8+ T cells. At 18-month median follow-up, patients with vaccine-expanded T cells (responders) had a longer median recurrence-free survival (not reached) compared with patients without vaccine-expanded T cells (non-responders; 13.4 months, P = 0.003). Differences in the immune fitness of the patients did not confound this correlation, as responders and non-responders mounted equivalent immunity to a concurrent unrelated mRNA vaccine against SARS-CoV-2. Thus, adjuvant atezolizumab, autogene cevumeran and mFOLFIRINOX induces substantial T cell activity that may correlate with delayed PDAC recurrence.

Fundamental immune-oncogenicity trade-offs define driver mutation fitness.

Missense driver mutations in cancer are concentrated in a few hotspots1. Various mechanisms have been proposed to explain this skew, including biased mutational processes2, phenotypic differences3-6 and immunoediting of neoantigens7,8; however, to our knowledge, no existing model weighs the relative contribution of these features to tumour evolution. We propose a unified theoretical 'free fitness' framework that parsimoniously integrates multimodal genomic, epigenetic, transcriptomic and proteomic data into a biophysical model of the rate-limiting processes underlying the fitness advantage conferred on cancer cells by driver gene mutations. Focusing on TP53, the most mutated gene in cancer1, we present an inference of mutant p53 concentration and demonstrate that TP53 hotspot mutations optimally solve an evolutionary trade-off between oncogenic potential and neoantigen immunogenicity. Our model anticipates patient survival in The Cancer Genome Atlas and patients with lung cancer treated with immunotherapy as well as the age of tumour onset in germline carriers of TP53 variants. The predicted differential immunogenicity between hotspot mutations was validated experimentally in patients with cancer and in a unique large dataset of healthy individuals. Our data indicate that immune selective pressure on TP53 mutations has a smaller role in non-cancerous lesions than in tumours, suggesting that targeted immunotherapy may offer an early prophylactic opportunity for the former. Determining the relative contribution of immunogenicity and oncogenic function to the selective advantage of hotspot mutations thus has important implications for both precision immunotherapies and our understanding of tumour evolution.

Neoantigen quality predicts immunoediting in survivors of pancreatic cancer.

Cancer immunoediting1 is a hallmark of cancer2 that predicts that lymphocytes kill more immunogenic cancer cells to cause less immunogenic clones to dominate a population. Although proven in mice1,3, whether immunoediting occurs naturally in human cancers remains unclear. Here, to address this, we investigate how 70 human pancreatic cancers evolved over 10 years. We find that, despite having more time to accumulate mutations, rare long-term survivors of pancreatic cancer who have stronger T cell activity in primary tumours develop genetically less heterogeneous recurrent tumours with fewer immunogenic mutations (neoantigens). To quantify whether immunoediting underlies these observations, we infer that a neoantigen is immunogenic (high-quality) by two features-'non-selfness'  based on neoantigen similarity to known antigens4,5, and 'selfness'  based on the antigenic distance required for a neoantigen to differentially bind to the MHC or activate a T cell compared with its wild-type peptide. Using these features, we estimate cancer clone fitness as the aggregate cost of T cells recognizing high-quality neoantigens offset by gains from oncogenic mutations. With this model, we predict the clonal evolution of tumours to reveal that long-term survivors of pancreatic cancer develop recurrent tumours with fewer high-quality neoantigens. Thus, we submit evidence that that the human immune system naturally edits neoantigens. Furthermore, we present a model to predict how immune pressure induces cancer cell populations to evolve over time. More broadly, our results argue that the immune system fundamentally surveils host genetic changes to suppress cancer.

Population variability in the generation and selection of T-cell repertoires.

The diversity of T-cell receptor (TCR) repertoires is achieved by a combination of two intrinsically stochastic steps: random receptor generation by VDJ recombination, and selection based on the recognition of random self-peptides presented on the major histocompatibility complex. These processes lead to a large receptor variability within and between individuals. However, the characterization of the variability is hampered by the limited size of the sampled repertoires. We introduce a new software tool SONIA to facilitate inference of individual-specific computational models for the generation and selection of the TCR beta chain (TRB) from sequenced repertoires of 651 individuals, separating and quantifying the variability of the two processes of generation and selection in the population. We find not only that most of the variability is driven by the VDJ generation process, but there is a large degree of consistency between individuals with the inter-individual variance of repertoires being about ∼2% of the intra-individual variance. Known viral-specific TCRs follow the same generation and selection statistics as all TCRs.

Generative models of T-cell receptor sequences.

T-cell receptors (TCR) are key proteins of the adaptive immune system, generated randomly in each individual, whose diversity underlies our ability to recognize infections and malignancies. Modeling the distribution of TCR sequences is of key importance for immunology and medical applications. Here, we compare two inference methods trained on high-throughput sequencing data: a knowledge-guided approach, which accounts for the details of sequence generation, supplemented by a physics-inspired model of selection; and a knowledge-free variational autoencoder based on deep artificial neural networks. We show that the knowledge-guided model outperforms the deep network approach at predicting TCR probabilities, while being more interpretable, at a lower computational cost.

ILC2s amplify PD-1 blockade by activating tissue-specific cancer immunity.

Group 2 innate lymphoid cells (ILC2s) regulate inflammation and immunity in mammalian tissues1,2. Although ILC2s are found in cancers of these tissues3, their roles in cancer immunity and immunotherapy are unclear. Here we show that ILC2s infiltrate pancreatic ductal adenocarcinomas (PDACs) to activate tissue-specific tumour immunity. Interleukin-33 (IL33) activates tumour ILC2s (TILC2s) and CD8+ T cells in orthotopic pancreatic tumours but not heterotopic skin tumours in mice to restrict pancreas-specific tumour growth. Resting and activated TILC2s express the inhibitory checkpoint receptor PD-1. Antibody-mediated PD-1 blockade relieves ILC2 cell-intrinsic PD-1 inhibition to expand TILC2s, augment anti-tumour immunity, and enhance tumour control, identifying activated TILC2s as targets of anti-PD-1 immunotherapy. Finally, both PD-1+ TILC2s and PD-1+ T cells are present in most human PDACs. Our results identify ILC2s as anti-cancer immune cells for PDAC immunotherapy. More broadly, ILC2s emerge as tissue-specific enhancers of cancer immunity that amplify the efficacy of anti-PD-1 immunotherapy. As ILC2s and T cells co-exist in human cancers and share stimulatory and inhibitory pathways, immunotherapeutic strategies to collectively target anti-cancer ILC2s and T cells may be broadly applicable.

OLGA: fast computation of generation probabilities of B- and T-cell receptor amino acid sequences and motifs.

MOTIVATION: High-throughput sequencing of large immune repertoires has enabled the development of methods to predict the probability of generation by V(D)J recombination of T- and B-cell receptors of any specific nucleotide sequence. These generation probabilities are very non-homogeneous, ranging over 20 orders of magnitude in real repertoires. Since the function of a receptor really depends on its protein sequence, it is important to be able to predict this probability of generation at the amino acid level. However, brute-force summation over all the nucleotide sequences with the correct amino acid translation is computationally intractable. The purpose of this paper is to present a solution to this problem. RESULTS: We use dynamic programming to construct an efficient and flexible algorithm, called OLGA (Optimized Likelihood estimate of immunoGlobulin Amino-acid sequences), for calculating the probability of generating a given CDR3 amino acid sequence or motif, with or without V/J restriction, as a result of V(D)J recombination in B or T cells. We apply it to databases of epitope-specific T-cell receptors to evaluate the probability that a typical human subject will possess T cells responsive to specific disease-associated epitopes. The model prediction shows an excellent agreement with published data. We suggest that OLGA may be a useful tool to guide vaccine design. AVAILABILITY AND IMPLEMENTATION: Source code is available at https://github.com/zsethna/OLGA. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Predicting the spectrum of TCR repertoire sharing with a data-driven model of recombination.

Despite the extreme diversity of T-cell repertoires, many identical T-cell receptor (TCR) sequences are found in a large number of individual mice and humans. These widely shared sequences, often referred to as "public," have been suggested to be over-represented due to their potential immune functionality or their ease of generation by V(D)J recombination. Here, we show that even for large cohorts, the observed degree of sharing of TCR sequences between individuals is well predicted by a model accounting for the known quantitative statistical biases in the generation process, together with a simple model of thymic selection. Whether a sequence is shared by many individuals is predicted to depend on the number of queried individuals and the sampling depth, as well as on the sequence itself, in agreement with the data. We introduce the degree of publicness conditional on the queried cohort size and the size of the sampled repertoires. Based on these observations, we propose a public/private sequence classifier, "PUBLIC" (Public Universal Binary Likelihood Inference Classifier), based on the generation probability, which performs very well even for small cohort sizes.

Insights into immune system development and function from mouse T-cell repertoires.

The ability of the adaptive immune system to respond to arbitrary pathogens stems from the broad diversity of immune cell surface receptors. This diversity originates in a stochastic DNA editing process (VDJ recombination) that acts on the surface receptor gene each time a new immune cell is created from a stem cell. By analyzing T-cell receptor (TCR) sequence repertoires taken from the blood and thymus of mice of different ages, we quantify the changes in the VDJ recombination process that occur from embryo to young adult. We find a rapid increase with age in the number of random insertions and a dramatic increase in diversity. Because the blood accumulates thymic output over time, blood repertoires are mixtures of different statistical recombination processes, and we unravel the mixture statistics to obtain a picture of the time evolution of the early immune system. Sequence repertoire analysis also allows us to detect the statistical impact of selection on the output of the VDJ recombination process. The effects we find are nearly identical between thymus and blood, suggesting that our analysis mainly detects selection for proper folding of the TCR receptor protein. We further find that selection is weaker in laboratory mice than in humans and it does not affect the diversity of the repertoire.

Inferring processes underlying B-cell repertoire diversity.

We quantify the VDJ recombination and somatic hypermutation processes in human B cells using probabilistic inference methods on high-throughput DNA sequence repertoires of human B-cell receptor heavy chains. Our analysis captures the statistical properties of the naive repertoire, first after its initial generation via VDJ recombination and then after selection for functionality. We also infer statistical properties of the somatic hypermutation machinery (exclusive of subsequent effects of selection). Our main results are the following: the B-cell repertoire is substantially more diverse than T-cell repertoires, owing to longer junctional insertions; sequences that pass initial selection are distinguished by having a higher probability of being generated in a VDJ recombination event; somatic hypermutations have a non-uniform distribution along the V gene that is well explained by an independent site model for the sequence context around the hypermutation site.

Accelerating multidimensional NMR and MRI experiments using iterated maps.

Techniques that accelerate data acquisition without sacrificing the advantages of fast Fourier transform (FFT) reconstruction could benefit a wide variety of magnetic resonance experiments. Here we discuss an approach for reconstructing multidimensional nuclear magnetic resonance (NMR) spectra and MR images from sparsely-sampled time domain data, by way of iterated maps. This method exploits the computational speed of the FFT algorithm and is done in a deterministic way, by reformulating any a priori knowledge or constraints into projections, and then iterating. In this paper we explain the motivation behind this approach, the formulation of the specific projections, the benefits of using a 'QUasi-Even Sampling, plus jiTter' (QUEST) sampling schedule, and various methods for handling noise. Applying the iterated maps method to real 2D NMR and 3D MRI of solids data, we show that it is flexible and robust enough to handle large data sets with significant noise and artifacts.