“Our goal is to put hard numbers on soft matter,” says Ernst-Ludwig Florin, a researcher at the University of Texas at Austin in the US, who led this latest research. His approach quantitatively images structures from the negative space excluded in the thermal random walk of a particle in an optical trap, monitored with laser light. With the technique he and his colleagues successfully imaged a collagen network with nanoscale resolution – despite its optical density – as well as quantifying the mechanical properties of the structure’s transverse fluctuations for the first time.

Florin began by studying single molecules with atomic force microscopy (AFM), where the topography of a surface is felt with atomic precision using a tip and cantilever in a similar way to a stylus reading a vinyl record. “AFM is limited to surfaces, so the idea was why not take optical tweezers and replace the tip with a particle and probe the cell from the inside,” explains Florin.

The problem was the high resolution of AFM results from the cantilever stiffness. In comparison, the optical forces on a particle in a trap are weak, leaving it subject to Brownian motion.

“We calculated how far the particle’s position fluctuates and that killed the idea of achieving super-resolution in and around cells,” says Florin. “But then I thought, why not let the particle explore a large space and we look at the excluded volume, and that gives the structure?”

Covering large volumes with Brownian motion

Florin worked with colleagues at the University of Texas at Austin, US, and Alexander University of Erlangen in Germany. Tobias Bartsch, the study’s first author, is now a postdoctoral fellow at Rockefeller University.

They used a polystyrene sphere with a diameter of around 190 nm ± 30 nm. The collagen network comprises filaments with diameters similar to the probe particle, and is mostly empty space. Left to Brownian motion alone the particle would be unlikely to find the network at all.

Instead the researchers developed a feedback system and scanned the network. They split each cubic micrometre into a 10 × 10 × 2 grid and focused the trap on each voxel in turn. This way the particle could explore smaller areas that could then be joined up.

LIGO lasers from astro to nano

Scanning large areas in this way requires an incredibly stable laser. After failed attempts with standard laser vendors the team turned to the lasers used in the LIGO experiments.

“We tested the LIGO laser and found it was only just stable enough,” says Florin. In some ways optical trap experiments are more demanding than LIGO, he tells nanotechweb.org. LIGO operates at a chosen frequency band that has low noise, but to track the position of the particle in the trap a much wider band (≈0.001Hz – 2MHz) needs to be integrated.

“That said, trapping experiments do not reach the sensitivity of the LIGO experiment, not even remotely,” adds Florin. The requirements of gravitational wave detection as a whole are so extreme that the system designed around the laser improves its frequency and power stability by a factor of 100 million for the LIGO experiments.

Quantifying assumptions

In addition to their low fill factor, collagen networks are optically dense and the scattering and refraction distorts the laser signal used to determine the position of the particle. To get around this, the researchers recorded the signal from the network with no particle in it and subtracted this from their measurements. “It’s simple and it worked,” says Florin. “We were quite surprised by that.”

Correcting for the distortion revealed vibrations in the collagen structure that the researchers could then quantify. Collagen is quite stiff so it has always been assumed that these vibrations are too small to contribute to the elastic properties of the network, but these are the first quantitative measurements to confirm this assumption. The researchers also imaged softer microtubules to show the efficacy of thermal noise imaging on softer structures.

Next, the team will look further into the mechanical information they can obtain from the probability function of the particle’s position.

Full details are reported in Nature Communications.