Monitoring the molecular composition of live cells exposed to electric pulses via label-free optical methods

The permeabilization of the live cells membrane by the delivery of electric pulses has fundamental interest in medicine, in particular in tumors treatment by electrochemotherapy. Since underlying mechanisms are still not fully understood, we studied the impact of electric pulses on the biochemical composition of live cells thanks to label-free optical methods: confocal Raman microspectroscopy and terahertz microscopy. A dose effect was observed after cells exposure to different field intensities and a major impact on cell peptide/protein content was found. Raman measurements reveal that protein structure and/or environment are modified by the electric pulses while terahertz measurements suggest a leakage of proteins and other intracellular compounds. We show that Raman and terahertz modalities are a particularly attractive complement to fluorescence microscopy which is the reference optical technique in the case of electropermeabilization. Finally, we propose an analytical model for the influx and efflux of non-permeant molecules through transiently (electro)permeabilized cell membranes.


Origin of the terahertz signal
The recorded terahertz signal variations originate from changes of the cytosol molecules concentration.
More precisely, the THz relative signal difference between the cells and their outer medium is proportional to the mass concentration of all intracellular molecules, from ions, metabolites to proteins.
To demonstrate this, we measured the THz relative signal obtained between simple solutions of molecules compared to pure water. Acquisitions of the THz relative signal were made on amino-acids, peptides and proteins up to 250 kDa. As shown in Figure S1a for a few examples, there was a linear relationship between a molecule mass concentration (in the biological range) and the THz relative signal generated. These slopes values are the THz relative signal sensitivity to mass concentration for each molecule. Values of all THz relative signal sensitivity given the size of the molecules tested are given in (in Figure S1b). The observed evolution of THz sensitivity with molecular weight can be understood and fitted by a simple volume model, as detailed in [1]. Figure S1: THz relative signal for amino-acid, peptide or protein solutions shows a linear relationship with mass concentration (a). Each slope illustrates the THz relative signal sensitivity to mass concentration of a given molecule. When plotted with respect to the molecular weight (b, black squares), sensitivity shows a global tendency that can be well fitted by a theoretical model (b, red line) such as described in [1]. From the smallest amino-acid to molecules of 1 kDa, the signal sensitivity sharply increases, meaning that a THz measurement will reflect both concentration and molecular weight of very small metabolites in a complex fashion, biased towards higher molecular weight for a given molar concentration. Above 2 kDa, sensitivity reaches a constant value (within a 12% margin), meaning that a THz measurement will then only probe the protein mass concentration regardless of their molecular weight.

Transient permeabilization model
The cytoplasmic membrane is a biological barrier that separates the interior of the cell to the extracellular space. It controls the flux of molecules in and out of the cells, being selectively permeable to specific molecules. In electropermeabilization, the membrane permeability can be transiently increased by applying an electric field, so that non-permeant molecules can cross the membrane.
We develop here a model to explain the behaviour of , which reflects the characteristic time for the efflux of molecules from the cytosol through the transiently permeabilized membrane. We report here how this transient permeability model applies to molecules efflux from the cells inside or to molecules influx depending on the initial concentration conditions.
To model this transient permeabilization, we consider here a membrane of surface surrounding a cell of volume , and the net efflux of molecules of molecular weight through the membrane, given by Fick's first law and the solubility-diffusion model for permeability [2], as where is the diffusion constant in cytosol, the membrane thickness, = / the concentration inside the cell and the effective diffusion area fraction, (that is the fraction of the unit area through which diffusion is effective for a molecule of mass M and Stokes radius ). The effective diffusion area fraction varies from 1 (fully open membrane) to 0 (closed membrane) and it is usually several orders of magnitude smaller than 1. The variation of the number of molecules inside the cell is then is the time-varying transfer rate constant. Let the permeability = / , the transfer rate writes = where = / is a constant depending only on the geometry of the cell. For efflux from the cytosol, the number of molecules is given by where 0 = ( = 0) is the internal concentration in the unpermeabilized cell. On an equivalent basis, for influx of fluorescent molecules into the cell, the solution is given by where is the concentration of fluorescent molecules in the extracellular space, assuming an infinite reservoir of molecules.
An applied electric field transiently modifies the effective diffusion area fraction ( ), which depends on the size of the electropores and on the Stokes radius of the molecules , following a Renkin model for diffusion through pores (see Figure S2) [3]. In mammalian cells, is found to decrease exponentially after electropermeabilization with a time constant independent of [4], We then model the evolution of the molecules through the cytosol membrane, integrating Eqs. (3) and (4), using The diffusion is given by Stokes-Einstein equation, where is the viscosity of the cytosol, the solute density, close to 1.37 g/cm 3 for a wide range of solute molecular weight [5] and NA the Avogadro's number. Additional parameters can be found in Table 1. The evolution of ( ) and ( ) are close to an exponential, and can be described by a characteristic decay time . Since depends on and , which both depend on and , we investigated the relative contributions of and on . We found (see Figure S3) that depends almost entirely on the contribution, an increase of the ratio / contributing to a decrease of , and very little on .