Research Briefing

Demultiplexing PET–MRI data of solid tumours using machine learning allows the spatial characterization of intratumour tissue heterogeneity in mice and humans. Predicted maps of tissue subtypes within the tumour could aid in conducting image-guided biopsies and provide valuable insights linking the outcome of cancer therapies with phenotypic heterogeneity.

We engineered integrase-deficient lentiviruses to act as vectors for the delivery of large gene knock-ins via homology-directed repair. This technology enables the non-cytotoxic, targeted insertion of difficult-to-express transgenes into genomic loci that are essential to cell survival, thereby overcoming the gene silencing that otherwise limits primary immune cell engineering.

We developed exponentially amplified rolling circle amplification with CRISPR–Cas12a as a one-pot, isothermal assay for microRNA detection. This method has single-digit femtomolar sensitivity and single-nucleotide specificity, and can be deployed for point-of-care testing. The assay has been adapted for the microRNA profiling of extracellular vesicles, which is used in the diagnosis of pancreatic cancer.

INSPECTR is a technique for detecting nucleic acids that couples the sensitivity and specificity of nucleic acid splinted ligation with the versatile readouts of cell-free gene expression. The result is an ambient-temperature workflow that enables the detection of pathogenic viruses at low copy numbers.

In this study, spherical nucleic acids (SNAs) were designed using defined placements of two classes of antigen that each activate different types of T cell. Use of SNA modularity illustrated the role of antigen placement within a vaccine architecture on subsequent processing and immune responses. The study defines design rules for multi-antigen vaccines.

We developed an at-home microsampling approach that measures thousands of metabolites, lipids and proteins in small volumes of blood. Dense multi-omic sampling generates a ‘molecular movie’ that integrates with data from wearables to reveal new insights into the dynamics of human physiology.

An extracellular matrix biomaterial delivered into the bloodstream selectively binds to blood vessels in inflamed tissues, such as those caused by myocardial infarction and traumatic brain injury. The biomaterial dampened the inflammatory response and promoted tissue repair and regeneration when tested in rat and pig models of myocardial infarction.

The inability to precisely manipulate mammalian mitochondrial DNA has stalled our understanding of mitochondrial biology and the generation of cellular and animal models in which to study it. DNA base editing technologies have enabled the generation of a library of mitochondrial base editors that precisely ablate every protein-coding gene in the mouse mitochondrial genome.

Graph deep learning applied to multiplexed immunofluorescence data from tumour microenvironments reveals spatial cellular structures that are indicative of cancer prognosis.

Developing gene therapy for use in the central nervous system has been hampered by the lack of an efficient vector for gene delivery. We report an adeno-associated virus vector with an enhanced ability to cross the blood–brain barrier in both rodents and non-human primates, and use it to develop systemic anti-tumour gene therapies for glioblastoma.

Multiplex detection of two interacting Mycobacterium tuberculosis biomarkers on the surface of circulating extracellular vesicles, using a nanoplasmon-enhanced immunoassay, improves the diagnosis of tuberculosis in immunosuppressed children living with HIV.

Graph deep learning can be used to detect contextual pathological features within a complex tumour microenvironment. We have shown the use of graph deep learning for predicting the prognosis of patients with tumours, and use it to identify additional contextual prognostic biomarkers for pathologists.

Ultrasound pulses have been used to modulate a liver–brain autonomic nerve pathway to prevent or reverse the onset of hyperglycaemia in models of diabetes in several species. The ion channel TRPA1 was shown to be essential in transducing the ultrasound stimuli within the metabolic control circuit.

A method connecting single-cell genomic, transcriptomic or proteomic profiles to functional cellular characteristics, especially time-varying phenotypic changes, would inform our understanding of cancer biology. We present functional single-cell sequencing (FUNseq) to address this need and describe how it might provide a unique way to unravel mechanisms that drive cancer.