An optofluidic chip for single-cell sorting based on mechanical properties

The past decade has seen an increasing interest in the analysis and sorting of single cells. Progress in this field has enabled researchers to overcome the limitations presented by heterogeneity in populations—found even in nominally identical cells—which poses severe challenges for many experimental measurements.¹ As a parameter for cell analysis, mechanical properties are attractive because they are sensitive markers of cell status, related to both pathological and physiological changes such as carcinogenesis,² differentiation, and proliferation.³ In contrast with alternative approaches, such as fluorescence-activated cell sorting, grouping on the basis of mechanical properties requires no specific marker to be exploited. There are several proposed techniques for measuring single-cell mechanical properties, including atomic force microscopy,⁴ micropipette aspiration,⁵ magnetic bead rheometry,⁶and optical tweezers.⁷ However, these methods have low throughput, and generally require direct contact with the cell, which may hamper its viability. While it is possible to achieve high throughput by purely hydrodynamic techniques,⁸ these only allow for single-cell mechanical analysis, without offering the possibility of sorting and extracting specific subpopulations.

To overcome all of these limitations, we may instead use a microfluidic optical stretcher, a technology that exploits optical forces to induce cell deformations, and which can be easily integrated in a microfluidic device. In doing so, we create an efficient tool to investigate cellular mechanical properties at the single-cell level.

An optical stretcher is typically built by placing two optical fibers facing one other, so as to obtain a dual-beam laser trap. We deliver the cells to the trapping region through a microfluidic channel that is designed to guarantee a controlled cell flow. Once a cell is firmly trapped, we achieve stretching by increasing the laser power at the fiber output.⁹

We realized a microfluidic stretcher with sorting capability driven by a cell's mechanical properties. We used femtosecond laser micromachining (FLM),¹⁰ a technology that enables integration of microfluidic and optical functions on the same glass chip, rapid 3D prototyping, and inherently precise alignment between optical and fluidic components. FLM is a two-step fabrication process. First, we realize permanent material modification through the nonlinear absorption of focused femtosecond laser pulses. Second, we etch the laser-modified zone using a hydrofluoric acid (HF) solution. Our microfluidic optical stretcher is composed of optical waveguides perpendicular to a microfluidic channel. By adding a Y-shaped bifurcation at the channel output, we exploit the optical forces to stretch the cell and also to push it toward the stream of one of the two outputs: see Figure 1(a). To limit roughness, we fabricated the channels parallel to the writing beam by irradiating them from the bottom to the top of the substrate, as shown schematically in Figure 1(b).¹¹ We realized the central channel, with a square cross section, by irradiating overlapped squares at different depths. The laser power and scanning velocity were varied along the substrate depth to overcome the effects of spherical aberrations. We obtained the Y-shaped branches by multiscan irradiation of lines to produce a channel the same height as the central one. We created access holes for the tubing connection by irradiating four coaxial circular helixes at each termination of the Y-branches. The irradiation pattern is shown in Figure 1(b).

Figure 1. (a) Working principle of the single-cell stretcher and sorter. Optical forces exerted by two optical waveguides facing the microchannel stretch cells and push them toward the desired output on the basis of their deformability. (b) Schematic rendering of the double-Y-shaped microfluidic circuit irradiation pattern. Fabrication of the device is by femtosecond laser micromachining. (c) The final device obtained after chemical etching.

In the same irradiation step, we also fabricated the optical waveguides for the stretching and sorting functionalities. After irradiation, we immersed the glass substrate in a 20% HF aqueous solution in an ultrasonic bath at 35°C for 1.5h. The final result is shown in Figure 1(c). We tested the device with an experiment aimed at a selection of cells from a suspension containing two different cellular lines of human melanoma: metastatic A375-P, and highly metastatic A365-MC2. When injected into the device, this suspension presented an equal concentration of the two cell lines. Each showed the same cellular size, but different deformability, as previously verified through an optical stretcher: see Figure 2(a). We chose a specific value of deformation as a threshold for selection, enabling separation of the population into two groups. In this specific case, we obtained an enriched A375-MC2 subpopulation by stretching all the cells and collecting only those with deformability higher than the threshold: see Figure 2(b).

Figure 2. (a) Normalized cellular distribution of human melanoma cellular lines A375-P and A375-MC2 as a function of their optical deformation. As an example, an 11% threshold is depicted and the colored areas show the number of cells expected in the collected sample for this threshold deformation value. (b) Percentage of A375-MC2 in the output of the sorted cells as a function of deformation threshold. Experimental points are reported together with the theoretical curve. Error bars are standard deviations.

In summary, we have proposed an integrated device that achieves stretching, sorting, and collection of single cells on the basis of their mechanical properties. This enables a new set of biological investigations, ranging from the sorting, separation, and analysis of cells with different mechanical properties, to the selection and recovery of cells showing specific mechanical response to drug treatments. Our future aim is the exploitation of this device as a building block in a complex optofluidic platform for single-cell analyses.

The authors are grateful for support from the Cariplo Agency, and colleagues from the University of Pavia who performed the biological experiments.