How to get into that "room at the bottom".

“There’s plenty of room at the bottom,” as Richard Feynman famously formulated it in 1959 (1). He considered the problem of manipulating and controlling things “at the bottom” of the length scale, from Ångströms to micrometers. The fame came retroactively with the rise of nanotechnology (2). By then, the advantage of studying small objects in small laboratories was clear: in a microscopic laboratory space, one can do things that are impossible in larger laboratories, such as inspecting a long DNA molecule visually (Fig. 1). In a larger volume of fluid, Brownian motion makes such a long, flexible molecule tangle up in a bundle, unless one pulls at its ends, which requires handles of a kind. It is done, but unpractical for inspection of many molecules, as they all would need handles. It seems simpler to flow the DNA molecule into a nanochannel so narrow that the molecule is fully stretched. Unfortunately, this is not simple, because there’s plenty of friction in that room at the bottom and an entropic barrier at its doorstep. This is a bottleneck for a promising complement to state-of-the-art DNA sequencing technologies. They could use the coarse-grained map of genomes in Fig. 1 to guide the assembly of sequenced DNA fragments (ref. 3, section 6.1). This bottleneck is the reason why the denaturation maps shown in Fig. 1 were produced in a nanoslit requiring human dexterity and not in a nanochannel. The effectively 2D slit is a compromise between the 3D volume, in which DNA bundles up, and the effectively 1D nanochannel that automatically stretches the DNA molecule. The slit gives us what we want, but not in the way we want it: it proves the concept optimally, but the required dexterity is not readily automated and scaled by parallelization. In PNAS, Berard et al. (4) avoid this bottleneck altogether with their convex lens-induced nanoscale templating (CLINT), which is a clever extension of a single-molecule imaging technique called convex lens-induced confinement (CLIC) (5). Instead of threading DNA molecules into nanochannels, Berard et al. form the nanochannels around their intended contents. This is no small trick, considering the intended contents are stretched DNA molecules. The molecules must be first stretched out and only then enclosed in the nanochannels. A chicken-and-egg problem? The authors’ trick is to harness the very entropic forces that oppose attempts to thread a DNA molecule into a nanochannel from its end. These are the same forces as those making rubber bands and O-rings Fig. 1. Composite photograph of individual double-stranded DNA molecules fluorescently dyed in a sequence-specific manner (barcoding) (7) and flow-stretched for visual inspection (8). Green vertical lines show three such molecules, each about 2 Mbp long, stretched in nanoslits by opposing flows. The downward flows end in a microchannel (greenish, at bottom). Orange dashed lines show enlarged details of the stretched molecules: two barcode patterns of fluorescence and dark patches. The top orange line shows 16 repeats: an impossible counting task for established sequencing methods, but trivial here, done with a single molecule. Insertions, deletions, and inversions were also found (8). Blue chromosomes: a single molecule that looked interesting in this visual inspection was retrieved from the nanofluidic device and whole genome amplified for further analysis. The barcodes shown here were obtained in a nanoslit that is 85 nm deep. There, dsDNA is effectively confined to a 2D world and consequently less tangled than in bulk. Thus, an experimenter with sufficient nanofluidic dexterity can balance the molecule in opposing flows that will stretch part of it up to 98% of its fully stretched length and totally suppress its longitudinal Brownian motion and the motion blur it causes in images. Maybe this manual skill can be automatized and parallelized. Berard et al. (4) present CLINT, which may be a faster, easier shortcut to similar results in nanochannels. Reprinted with permission from ref. 8. Author contributions: H.F. wrote the paper.