An Automated Microfluidic Platform for In Vitro Raman Analysis of Living Cells

Abstract

We present a miniaturized, inexpensive, and user-friendly microfluidic platform to support biological applications. The system integrates a mini-incubator providing controlled environmental conditions and housing a microfluidic device for long-term cell culture experiments. The incubator is designed to be compatible with standard inverted optical microscopes and Raman spectrometers, allowing for the non-invasive imaging and spectroscopic analysis of cell cultures in vitro. The microfluidic device, which reproduces a dynamic environment, was optimized to sustain a passive, gravity-driven flow of medium, eliminating the need for an external pumping system and reducing mechanical stress on the cells. The platform was tested using Raman analysis and adherent tumoral cells to assess proliferation prior and subsequent to hydrogen peroxide treatment for oxidative stress induction. The results demonstrated a successful adhesion of cells onto the substrate and their proliferation. Furthermore, the platform is suitable for carrying out optical monitoring of cultures and Raman analysis. In fact, it was possible to discriminate spectra deriving from control and hydrogen peroxide-treated cells in terms of DNA backbone and cellular membrane modification effects provoked by reactive oxygen species (ROS) activity. The 800-1100 cm^-1 band highlights the destructive effects of ROS on the DNA backbone's structure, as its rupture modifies its vibration; moreover, unpaired nucleotides are increased in treated sample, as shown in the 1154-1185 cm^-1 band. Protein synthesis deterioration, led by DNA structure damage, is highlighted in the 1257-1341 cm^-1, 1440-1450 cm^-1, and 1640-1670 cm^-1 bands. Furthermore, membrane damage is emphasized in changes in the 1270, 1301, and 1738 cm^-1 frequencies, as phospholipid synthesis is accelerated in an attempt to compensate for the membrane damage brought about by the ROS attack. This study highlights the potential use of this platform as an alternative to conventional culturing and analysis procedures, considering that cell culturing, optical imaging, and Raman spectroscopy can be performed simultaneously on living cells with minimal cellular stress and without the need for labeling or fixation.

Keywords: Raman spectroscopy; cancer cells; cell culturing; in vitro culturing; microfluidic screening devices; microfluidics; mini-incubator; optical imaging.

Figure 1

Three-dimensional scheme of the PMD. (a) Representation of the three layers of the PMD: (i) a cover layer was machined on both faces; the internal face has openings that let the liquids pass through, and the external face has a large central optical window manufactured after the bonding process; (ii) a channel layer was machined on both faces; the bottom face has 5 microchannels, while the top face has a large optical central window manufactured after the bonding process; (iii) a bottom layer was machined on both faces; the upper face has a milled 2.7 mm deep tank which will create the collection tank (waste); at the center, there is an opening for cell culturing, which is closed on the bottom by attaching a CaF₂ slide; the slide is attached inside a milled pocket after the bonding process. It is worth highlighting that the bottom and the upper circular optical openings allow the microscope objective to approach the cell culture in both direct and inverted configurations. (b) Upper view of the device with its media input reservoirs A, B, C, D, waste reservoir, and the upper optical window. (c) Cross-sectional view of the device, namely the cells in the central reservoir, which have to be fed by the media carried by the microchannels; below the cell culture’s reservoir, there is a CaF₂ slide attached. (d) Cross-sectional detail of the CaF₂ slide’s positioning with respect to the microchannel plane. The adhesive tape is positioned between the bottom layer and the CaF₂ slide. (e) Schematic view of the cells lying on a plane 100 μm below the microchannel plane because of the adhesive tape’s thickness, denoted with h₁.

Figure 2

The microfluidic platform, consisting of the MI and PMD, is connected to the external dedicated PID controller, powered by an external generator, which monitors the cart heater temperature thanks to a K-type thermocouple. For CO₂ supply to the system, a cylinder with a gas mixture of 20% O₂, 5% CO₂, and 75% N₂ is used. The assembled MI was coupled to an inVia Raman spectrometer from Renishaw, Gloucestershire, UK, or to an Eclipse-Ti inverted microscope (Nikon Instruments Inc., Melville, NY, USA), depending on the experiments. An IR laser is utilized in this setup. A graphic user interface developed in Python 3.12 is used to monitor and set the parameters. A water level sensor is utilized to trace the water level in the MI’s internal humidity tank.

Figure 3

(left) The resulting microfluidic platform composed of the MI and PMD prior to its working conditions; (center) top view of the platform inside the MI along with the touch screen with dedicated electronics and GUI allowing for the setting and monitoring of incubation parameters; (right) close-up view of the PMD used in this work.

Figure 4

Experimental temperature (left) and humidity (right) profiles recorded for 2 h.

Figure 5

Microscope photograph of the resulting culture after 72 h (left); growth curve of HeLa cells during the PMD functioning assessment at different time periods(right).

Figure 6

The BF, PC1, and clustering images are shown for each sample. It can be seen how cellular compartments are well-clustered in the control sample, unlike in the treated cell. In the cluster maps, the white color is related to the background, while the green, yellow, and blue colors refer, respectively, to the external membrane, cytoplasm, and nuclear region. The latter seems to be very fragmented in the treated sample (top). Average spectrum calculated for each cluster, considering the signal coming from both the samples. Note how the signals associated with the nuclear and cytoplasm cluster are the most significant (bottom).

Figure 7

Average cellular spectra of the nucleus and cytoplasm (respectively, the yellow and blue classes in Figure 6) for the control and treated samples. The resulting spectra are smoothed and plotted together to perform a differential analysis. The difference spectrum is calculated by subtracting the treated sample from the control. The signal-to-noise ratio (SNR) is 2.87. Noise was evaluated on the difference spectrum, and it was compared with the most significant Raman bands. The resulting difference spectrum is amplified by a 2.5 factor.