Tumor-agnostic transcriptome-based classifier identifies spatial infiltration patterns of CD8+T cells in the tumor microenvironment and predicts clinical outcome in early-phase and late-phase clinical trials.

BACKGROUND: The immune status of a patient's tumor microenvironment (TME) may guide therapeutic interventions with cancer immunotherapy and help identify potential resistance mechanisms. Currently, patients' immune status is mostly classified based on CD8+tumor-infiltrating lymphocytes. An unmet need exists for comparable and reliable precision immunophenotyping tools that would facilitate clinical treatment-relevant decision-making and the understanding of how to overcome resistance mechanisms. METHODS: We systematically analyzed the CD8 immunophenotype of 2023 patients from 14 phase I-III clinical trials using immunohistochemistry (IHC) and additionally profiled gene expression by RNA-sequencing (RNA-seq). CD8 immunophenotypes were classified by pathologists into CD8-desert, CD8-excluded or CD8-inflamed tumors using CD8 IHC staining in epithelial and stromal areas of the tumor. Using regularized logistic regression, we developed an RNA-seq-based classifier as a surrogate to the IHC-based spatial classification of CD8+tumor-infiltrating lymphocytes in the TME. RESULTS: The CD8 immunophenotype and associated gene expression patterns varied across indications as well as across primary and metastatic lesions. Melanoma and kidney cancers were among the strongest inflamed indications, while CD8-desert phenotypes were most abundant in liver metastases across all tumor types. A good correspondence between the transcriptome and the IHC-based evaluation enabled us to develop a 92-gene classifier that accurately predicted the IHC-based CD8 immunophenotype in primary and metastatic samples (area under the curve inflamed=0.846; excluded=0.712; desert=0.855). The newly developed classifier was prognostic in The Cancer Genome Atlas (TCGA) data and predictive in lung cancer: patients with predicted CD8-inflamed tumors showed prolonged overall survival (OS) versus patients with CD8-desert tumors (HR 0.88; 95% CI 0.80 to 0.97) across TCGA, and longer OS on immune checkpoint inhibitor administration (phase III OAK study) in non-small-cell lung cancer (HR 0.75; 95% CI 0.58 to 0.97). CONCLUSIONS: We provide a new precision immunophenotyping tool based on gene expression that reflects the spatial infiltration patterns of CD8+ lymphocytes in tumors. The classifier enables multiplex analyses and is easy to apply for retrospective, reverse translation approaches as well as for prospective patient enrichment to optimize the response to cancer immunotherapy.

A first-in-human study of the fibroblast activation protein-targeted, 4-1BB agonist RO7122290 in patients with advanced solid tumors.

This first-in-human study evaluated RO7122290, a bispecific fusion protein carrying a split trimeric 4-1BB (CD137) ligand and a fibroblast activation protein α (FAP) binding site that costimulates T cells for improved tumor cell killing in FAP-expressing tumors. Patients with advanced or metastatic solid tumors received escalating weekly intravenous doses of RO7122290 as a single agent (n = 65) or in combination with a 1200-milligram fixed dose of the anti-programmed death-ligand 1 (anti-PD-L1) antibody atezolizumab given every 3 weeks (n = 50), across a tested RO7122290 dose range of 5 to 2000 milligrams and 45 to 2000 milligrams, respectively. Three dose-limiting toxicities were reported, two at different RO7122290 single-agent doses (grade 3 febrile neutropenia and grade 3 cytokine release syndrome) and one for the combination (grade 3 pneumonitis). No maximum tolerated dose was identified. The pharmacokinetic profile of RO7122290 suggested nonlinearity in elimination. The observed changes in peripheral and tissue pharmacodynamic (PD) biomarkers were consistent with the postulated mechanism of action. Treatment-induced PD changes included an increase in proliferating and activated T cells in peripheral blood both in the single-agent and combination arms. Increased infiltration of intratumoral CD8+ and Ki67+CD8+ T cells was observed for both treatment regimens, accompanied by the up-regulation of T cell activation genes and gene signatures. Eleven patients experienced a complete or partial response, six of whom were confirmed to be immune checkpoint inhibitor naive. These results support further evaluation of RO7122290 in combination with atezolizumab or other immune-oncology agents for the treatment of solid tumors.

Deciphering molecular and cellular ex vivo responses to bispecific antibodies PD1-TIM3 and PD1-LAG3 in human tumors.

BACKGROUND: Next-generation cancer immunotherapies are designed to broaden the therapeutic repertoire by targeting new immune checkpoints including lymphocyte-activation gene 3 (LAG-3) and T cell immunoglobulin and mucin-domain containing-3 (TIM-3). Yet, the molecular and cellular mechanisms by which either receptor functions to mediate its inhibitory effects are still poorly understood. Similarly, little is known on the differential effects of dual, compared with single, checkpoint inhibition. METHODS: We here performed in-depth characterization, including multicolor flow cytometry, single cell RNA sequencing and multiplex supernatant analysis, using tumor single cell suspensions from patients with cancer treated ex vivo with novel bispecific antibodies targeting programmed cell death protein 1 (PD-1) and TIM-3 (PD1-TIM3), PD-1 and LAG-3 (PD1-LAG3), or with anti-PD-1. RESULTS: We identified patient samples which were responsive to PD1-TIM3, PD1-LAG3 or anti-PD-1 using an in vitro approach, validated by the analysis of 659 soluble proteins and enrichment for an anti-PD-1 responder signature. We found increased abundance of an activated (HLA-DR+CD25+GranzymeB+) CD8+ T cell subset and of proliferating CD8+ T cells, in response to bispecific antibody or anti-PD-1 treatment. Bispecific antibodies, but not anti-PD-1, significantly increased the abundance of a proliferating natural killer cell subset, which exhibited enrichment for a tissue-residency signature. Key phenotypic and transcriptional changes occurred in a PD-1+CXCL13+CD4+ T cell subset, in response to all treatments, including increased interleukin-17 secretion and signaling toward plasma cells. Interestingly, LAG-3 protein upregulation was detected as a unique pharmacodynamic effect mediated by PD1-LAG3, but not by PD1-TIM3 or anti-PD-1. CONCLUSIONS: Our in vitro system reliably assessed responses to bispecific antibodies co-targeting PD-1 together with LAG-3 or TIM-3 using patients' tumor infiltrating immune cells and revealed transcriptional and phenotypic imprinting by bispecific antibody formats currently tested in early clinical trials.

An Introduction to Machine Learning.

In the last few years, machine learning (ML) and artificial intelligence have seen a new wave of publicity fueled by the huge and ever-increasing amount of data and computational power as well as the discovery of improved learning algorithms. However, the idea of a computer learning some abstract concept from data and applying them to yet unseen situations is not new and has been around at least since the 1950s. Many of these basic principles are very familiar to the pharmacometrics and clinical pharmacology community. In this paper, we want to introduce the foundational ideas of ML to this community such that readers obtain the essential tools they need to understand publications on the topic. Although we will not go into the very details and theoretical background, we aim to point readers to relevant literature and put applications of ML in molecular biology as well as the fields of pharmacometrics and clinical pharmacology into perspective.

Duplication history and molecular evolution of the rbcS multigene family in angiosperms.

Ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) is considered to be the main enzyme determining the rate of photosynthesis. The small subunit of the protein, encoded by the rbcS gene, has been shown to influence the catalytic efficiency, CO2 specificity, assembly, activity, and stability of RuBisCO. However, the evolution of the rbcS gene remains poorly studied. We inferred the phylogenetic tree of the rbcS gene in angiosperms using the nucleotide sequences and found that it is composed of two lineages that may have existed before the divergence of land plants. Although almost all species sampled carry at least one copy of lineage 1, genes of lineage 2 were lost in most angiosperm species. We found the specific residues that have undergone positive selection during the evolution of the rbcS gene. We detected intensive coevolution between each rbcS gene copy and the rbcL gene encoding the large subunit of RuBisCO. We tested the role played by each rbcS gene copy on the stability of the RuBisCO protein through homology modelling. Our results showed that this evolutionary constraint could limit the level of divergence seen in the rbcS gene, which leads to the similarity among the rbcS gene copies of lineage 1 within species.

Large-Scale Comparative Analysis of Codon Models Accounting for Protein and Nucleotide Selection.

There are numerous sources of variation in the rate of synonymous substitutions inside genes, such as direct selection on the nucleotide sequence, or mutation rate variation. Yet scans for positive selection rely on codon models which incorporate an assumption of effectively neutral synonymous substitution rate, constant between sites of each gene. Here we perform a large-scale comparison of approaches which incorporate codon substitution rate variation and propose our own simple yet effective modification of existing models. We find strong effects of substitution rate variation on positive selection inference. More than 70% of the genes detected by the classical branch-site model are presumably false positives caused by the incorrect assumption of uniform synonymous substitution rate. We propose a new model which is strongly favored by the data while remaining computationally tractable. With the new model we can capture signatures of nucleotide level selection acting on translation initiation and on splicing sites within the coding region. Finally, we show that rate variation is highest in the highly recombining regions, and we propose that recombination and mutation rate variation, such as high CpG mutation rate, are the two main sources of nucleotide rate variation. Although we detect fewer genes under positive selection in Drosophila than without rate variation, the genes which we detect contain a stronger signal of adaptation of dynein, which could be associated with Wolbachia infection. We provide software to perform positive selection analysis using the new model.

Genome rearrangements and selection in multi-chromosome bacteria Burkholderia spp.

BACKGROUND: The genus Burkholderia consists of species that occupy remarkably diverse ecological niches. Its best known members are important pathogens, B. mallei and B. pseudomallei, which cause glanders and melioidosis, respectively. Burkholderia genomes are unusual due to their multichromosomal organization, generally comprised of 2-3 chromosomes. RESULTS: We performed integrated genomic analysis of 127 Burkholderia strains. The pan-genome is open with the saturation to be reached between 86,000 and 88,000 genes. The reconstructed rearrangements indicate a strong avoidance of intra-replichore inversions that is likely caused by selection against the transfer of large groups of genes between the leading and the lagging strands. Translocated genes also tend to retain their position in the leading or the lagging strand, and this selection is stronger for large syntenies. Integrated reconstruction of chromosome rearrangements in the context of strains phylogeny reveals parallel rearrangements that may indicate inversion-based phase variation and integration of new genomic islands. In particular, we detected parallel inversions in the second chromosomes of B. pseudomallei with breakpoints formed by genes encoding membrane components of multidrug resistance complex, that may be linked to a phase variation mechanism. Two genomic islands, spreading horizontally between chromosomes, were detected in the B. cepacia group. CONCLUSIONS: This study demonstrates the power of integrated analysis of pan-genomes, chromosome rearrangements, and selection regimes. Non-random inversion patterns indicate selective pressure, inversions are particularly frequent in a recent pathogen B. mallei, and, together with periods of positive selection at other branches, may indicate adaptation to new niches. One such adaptation could be a possible phase variation mechanism in B. pseudomallei.

State aggregation for fast likelihood computations in molecular evolution.

Motivation: Codon models are widely used to identify the signature of selection at the molecular level and to test for changes in selective pressure during the evolution of genes encoding proteins. The large size of the state space of the Markov processes used to model codon evolution makes it difficult to use these models with large biological datasets. We propose here to use state aggregation to reduce the state space of codon models and, thus, improve the computational performance of likelihood estimation on these models. Results: We show that this heuristic speeds up the computations of the M0 and branch-site models up to 6.8 times. We also show through simulations that state aggregation does not introduce a detectable bias. We analyzed a real dataset and show that aggregation provides highly correlated predictions compared to the full likelihood computations. Finally, state aggregation is a very general approach and can be applied to any continuous-time Markov process-based model with large state space, such as amino acid and coevolution models. We therefore discuss different ways to apply state aggregation to Markov models used in phylogenetics. Availability and Implementation: The heuristic is implemented in the godon package (https://bitbucket.org/Davydov/godon) and in a version of FastCodeML (https://gitlab.isb-sib.ch/phylo/fastcodeml). Contact: nicolas.salamin@unil.ch Supplementary Information: Supplementary data are available at Bioinformatics online.

Energy barriers and driving forces in tRNA translocation through the ribosome.

During protein synthesis, tRNAs move from the ribosome's aminoacyl to peptidyl to exit sites. Here we investigate conformational motions during spontaneous translocation, using molecular dynamics simulations of 13 intermediate-translocation-state models obtained by combining Escherichia coli ribosome crystal structures with cryo-EM data. Resolving fast transitions between states, we find that tRNA motions govern the transition rates within the pre- and post-translocation states. Intersubunit rotations and L1-stalk motion exhibit fast intrinsic submicrosecond dynamics. The L1 stalk drives the tRNA from the peptidyl site and links intersubunit rotation to translocation. Displacement of tRNAs is controlled by 'sliding' and 'stepping' mechanisms involving conserved L16, L5 and L1 residues, thus ensuring binding to the ribosome despite large-scale tRNA movement. Our results complement structural data with a time axis, intrinsic transition rates and molecular forces, revealing correlated functional motions inaccessible by other means.

Evolution of the protein stoichiometry in the L12 stalk of bacterial and organellar ribosomes.

The emergence of ribosomes and translation factors is central for understanding the origin of life. Recruitment of translation factors to bacterial ribosomes is mediated by the L12 stalk composed of protein L10 and several copies of protein L12, the only multi-copy protein of the ribosome. Here we predict stoichiometries of L12 stalk for >1,200 bacteria, mitochondria and chloroplasts by a computational analysis, and validate the predictions by quantitative mass spectrometry. The majority of bacteria have L12 stalks allowing for binding of four or six copies of L12, largely independent of the taxonomic group or living conditions of the bacteria, whereas some cyanobacteria have eight copies. Mitochondrial and chloroplast ribosomes can accommodate six copies of L12. The last universal common ancestor probably had six molecules of L12 molecules bound to L10. Changes of the stalk composition provide a unique possibility to trace the evolution of protein components of the ribosome.