Gene machines

Abstract:

Transcription in bacteria is controlled by multiple molecular mechanisms that precisely regulate gene expression. It has been recently shown that initial RNA synthesis by the bacterial RNA polymerase (RNAP) is interrupted by pauses; however, the pausing determinants and the relationship of pausing with productive and abortive RNA synthesis remain poorly understood. Using single-molecule FRET and biochemical analysis, here we show that the pause encountered by RNAP after the synthesis of a 6-nt RNA (ITC6) renders the promoter escape strongly dependent on the NTP concentration. Mechanistically, the paused ITC6 acts as a checkpoint that directs RNAP to one of three competing pathways: productive transcription, abortive RNA release, or a new unscrunching/scrunching pathway. The cyclic unscrunching/scrunching of the promoter generates a long-lived, RNA-bound paused state; the abortive RNA release and DNA unscrunching are thus not as tightly linked as previously thought. Finally, our new model couples the pausing with the abortive and productive outcomes of initial transcription.

Abstract:

DNA sequence regulates complex reactions and protein-DNA interactions, yet sequence effects remain poorly understood due to the lack of direct, high-throughput approaches to study sequence-dependence at the single-molecule level. Here, we introduce Single-molecule Phenotyping and In-Situ Sequencing (SPIN-Seq), a DNA-based, protein-free method that links functional and structural properties of a single DNA molecule with the sequence of that same molecule. After performing functional assays on immobilized DNA molecules, SPIN-Seq uses sequencing-by-transient-hybridization on the same surface and instrument to read sequences in each immobilized DNA molecule, directly linking phenotype and genotype. Applying SPIN-Seq, we dissected the interaction of a transcription factor with its target sequence, and revealed the sequence-dependence of RNA polymerase pausing and reaction-path branching during initial transcription. Our method provides powerful, systematic ways to understand complex molecular mechanisms and their sequence dependence.

Abstract:

The convergence of biosensing and nucleic acid (NA) nanotechnology represents an opportunity for the development of diagnostic technologies. By harnessing the programmability of nucleic acids, we can design biosensors that offer advantages in stability, scalability, versatility and sensitivity, compared to protein-based systems. In this work we introduce DNA-FLASH (DNA-based FLuorescence Amplification upon Single-target Hybridization), a DNA nanosensor concept for digital biosensing. DNA-FLASH leverages fluorescence amplification by a multicomponent NA enzyme (MNAzyme)-driven DNA walker mechanism on a DNA origami disk. Using super-resolution microscopy and single-molecule photobleaching, we demonstrate reproducible fabrication of DNA-FLASH nanosensors with 12 fluorophore-quencher substrates on a ring-shaped track, surrounding a single MNAzyme walker. This nanoarchitecture enables single-molecule detection of DNA targets down to picomolar concentrations. Through precise patterning of DNA-FLASH nanosensors in arrays on glass, we facilitate high-throughput single-molecule readout. We successfully demonstrate DNA-FLASH in human plasma samples and on an in-house developed, fully integrated, self-powered, disposable microfluidic chip, highlighting its potential use in point-of-care settings. Altogether, DNA-FLASH may support the development of next-generation biosensors capable of addressing pressing global challenges, including rapid disease detection, environmental sustainability, and personalized healthcare.

Abstract:

Single-molecule fluorescence resonance energy transfer (smFRET) is a versatile technique for studying biomolecular dynamics and function by detecting nanoscale movements as fluorescence signals. Analysing such signals is a complex exercise, which has recently been the focus of approaches relying on deep learning. Here, we survey such artificial-intelligence-based approaches and compare them with classical methods for smFRET analysis. The use of deep learning has shown potential to enhance precision, accuracy, and speed in analysing massive smFRET datasets.